tpch2 8 0

TPC BENCHMARKTM H
(Decision Support)
Standard Specification
Revision 2.8.0
Transaction Processing Performance Council (TPC)

Presidio of San Francisco
Building 572B Ruger St. (surface)
P.O. Box 29920 (mail)
San Francisco, CA 94129-0920
Voice:415-561-6272
Fax:415-561-6120
Email: [email protected]
1993 - 2008 Transaction Processing Performance Council
TPC Benchmark H Standard Specification Revision 2.8.0
Page 1
Acknowledgments
The TPC acknowledges the work and contributions of the TPC-D subcommittee member companies in developing
Version 2 of the TPC-D specification which formed the basis for TPC-H Version 1. The subcommittee included representatives from Compaq, Data General, Dell, EMC, HP, IBM, Informix, Microsoft, NCR, Oracle, Sequent, SGI,
Sun, Sybase, and Unisys. The TPC also acknowledges the contribution of Jack Stephens, consultant to the TPC-D
subcommittee, for his work on the benchmark specification and DBGEN development.
TPC Membership
(as of September 23, 2008)
Hitachi SW
IBM Corp.
Ingres
Intel Corporation.
Kickfire
Microsoft Corporation.
NEC
Netezza
AMD
Bull S.A.
Dell Inc.
Exasol
Fujitsu
Fusion IO
Greenplum
Hewlett-Packard
Oracle Corporation
ParAccel
Fujitsu-Siemens
Sun Microsystems
Sybase
Teradata
Unisys Corporation
Vertica
Vmware
Document History
Date
Version
Description
26 February 1999
Draft 1.0.0
Mail ballot draft for Standard Specification
24 June 1999
Revision 1.1.0
First minor revision of the Specification
25 April 2002
Revision 1.4.0
Clarification about Primary Keys
12 July 2002
Revision 1.5.0
Additions for EOL of hardware in 8.6
15 July 2002
Revision 2.0.0
Mail ballot draft 3 year maintenance pricing
14 August 2003
Revision 2.1.0
Adding scale factors 30TB and 100TB
29 June 2005
Revision 2.2.0
Adding Pricing Specification 1.0.0
11 August 2005
Revision 2.3.0
Changing pricing precision to cents and processor definition
Page 2
Date
Version
Description
23 June 2006
Revision 2.4.0
Adding reference data set and audit requirements to verify populated database, effect of
update data and qgen substitution parameters.
Scale factors larger than 10,000 are required to
use this version.
10 July 2006
Revision 2.5.0
dbgen bug fixes in parallel data generation,

updates to reference data set/qualification output, modified audit rules and updated executive
summary example.
26 October 2006
Revision 2.6.0
Added Clause 7.2.3.1 about software license

pricing, removed Clause 7.1.3.3 about 8 hour
log requirement and updated executive summary example in Appendix E
14 June 2006
Revision 2.6.1
Editorial correction in Clause 2.1.3.3

Clarification of Clause 9.2.4.5
28 February 2008
Revision 2.6.2
Change substr into substring in Clause 2.25.2,

update of membership list, TPC address and
copyright statement
17 April 2008
Revision 2.7.0
Incorporate BUG fix 595 of qgen
11 September 2008
Revision 2.8.0
Add wording to allow substitutions in Clause

7.2. Modify clauses 5.4, 5.4.6, 8.4.2.2 and
9.2.6.1 to refer to pricing specification. Update
TPC member companies.
TPC Benchmark, TPC-H, QppH, QthH, and QphH are trademarks of the Transaction Processing Performance
Council.
All parties are granted permission to copy and distribute to any party without fee all or part of this material provided that: 1) copying and distribution is done for the primary purpose of disseminating TPC material; 2) the TPC
copyright notice, the title of the publication, and its date appear, and notice is given that copying is by permission of
the Transaction Processing Performance Council.
Parties wishing to copy and distribute TPC materials other than for the purposes outlined above (including incorporating TPC material in a nonTPC document, specification or report), must secure the TPC's written permission.
Page 3
Table of Contents
0
7
8
INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
0.1
Preamble . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
0.2
General Implementation Guidelines . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
0.3
General Measurement Guidelines. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
LOGICAL DATABASE DESIGN . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.1
Business and Application Environment . . . . . . . . . . . . . . . . . . . . . . . 10
1.2
Database Entities, Relationships, and Characteristics . . . . . . . . . . . . 12
1.3
Datatype Definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.4
Table Layouts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.5
Implementation Rules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
1.6
Data Access Transparency Requirements. . . . . . . . . . . . . . . . . . . . . . 20
QUERIES AND REFRESH FUNCTIONS . . . . . . . . . . . . . . . . . . . . . 21
2.1
General Requirements and Definitions for Queries . . . . . . . . . . . . . . 21
2.2
Query Compliance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.3
Query Validation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.4
New Sales Refresh Function (RF1) . . . . . . . . . . . . . . . . . . . . . . . . . . 74
2.5
Old Sales Refresh Function (RF2) . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
2.6
Database Evolution Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
DATABASE SYSTEM PROPERTIES. . . . . . . . . . . . . . . . . . . . . . . . . 77
3.1
The ACID Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
3.2
Atomicity Requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
3.3
Consistency Requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
3.4
Isolation Requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
3.5
Durability Requirements. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
SCALING AND DATABASE POPULATION . . . . . . . . . . . . . . . . . . 87
4.1
Database Definition and Scaling. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
4.2
DBGEN and Database Population . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
4.3
Database Load Time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
PERFORMANCE METRICS AND EXECUTION RULES . . . . . . 101
5.1
Definition of Terms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
5.2
Configuration Rules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
5.3
Execution Rules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
5.4
Metrics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
SUT AND DRIVER IMPLEMENTATION . . . . . . . . . . . . . . . . . . . . 112
6.1
Models of Tested Configurations . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
6.2
System Under Test (SUT) Definition . . . . . . . . . . . . . . . . . . . . . . . . 112
6.3
Driver Definition. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
PRICING . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
7.1
Priced System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
FULL DISCLOSURE. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
8.1
Reporting Requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
8.2
Format Guidelines. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
8.3
Full Disclosure Report Contents. . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
8.4
Executive Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
8.5
Availability of the Full Disclosure Report . . . . . . . . . . . . . . . . . . . . 128
Page 4
8.6
9
Revisions to the Full Disclosure Report . . . . . . . . . . . . . . . . . . . . . . 128

AUDIT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
9.1
General Rules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
9.2
Auditor's Check List . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
Appendix A ORDERED SETS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
Appendix B APPROVED QUERY VARIANTS . . . . . . . . . . . . . . . . . . 135
Appendix C QUERY VALIDATION . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
Appendix D DATA AND QUERY GENERATION PROGRAMS . . . 140
Appendix E
. . . . . . . . . . . . . . . SAMPLE EXECUTIVE SUMMARY141
Appendix F REFERENCE DATA SET . . . . . . . . . . . . . . . . . . . . . . . . . 149
Page 5
0: INTRODUCTION
0.1
Preamble
The TPC BenchmarkH (TPC-H) is a decision support benchmark. It consists of a suite of business oriented ad-hoc
queries and concurrent data modifications. The queries and the data populating the database have been chosen to
have broad industry-wide relevance while maintaining a sufficient degree of ease of implementation. This benchmark illustrates decision support systems that
Examine large volumes of data;
Execute queries with a high degree of complexity;
Give answers to critical business questions.
TPC-H evaluates the performance of various decision support systems by the execution of sets of queries against a
standard database under controlled conditions. The TPC-H queries:
Give answers to real-world business questions;
Simulate generated ad-hoc queries (e.g., via a point and click GUI interface);
Are far more complex than most OLTP transactions;
Include a rich breadth of operators and selectivity constraints;
Generate intensive activity on the part of the database server component of the system under test;
Are executed against a database complying to specific population and scaling requirements;
Are implemented with constraints derived from staying closely synchronized with an on-line production database.
The TPC-H operations are modeled as follows:
The database is continuously available 24 hours a day, 7 days a week, for ad-hoc queries from multiple end
users and data modifications against all tables, except possibly during infrequent (e.g., once a month) maintenance sessions;
The TPC-H database tracks, possibly with some delay, the state of the OLTP database through on-going
refresh functions which batch together a number of modifications impacting some part of the decision support
database;
Due to the world-wide nature of the business data stored in the TPC-H database, the queries and the refresh
functions may be executed against the database at any time, especially in relation to each other. In addition,
this mix of queries and refresh functions is subject to specific ACIDity requirements, since queries and refresh
functions may execute concurrently;
To achieve the optimal compromise between performance and operational requirements, the database administrator can set, once and for all, the locking levels and the concurrent scheduling rules for queries and refresh
functions.
The minimum database required to run the benchmark holds business data from 10,000 suppliers. It contains almost
ten million rows representing a raw storage capacity of about 1 gigabyte. Compliant benchmark implementations
may also use one of the larger permissible database populations (e.g., 100 gigabytes), as defined in Clause 5.1.3.
The performance metric reported by TPC-H is called the TPC-H Composite Query-per-Hour Performance Metric
(QphH@Size), and reflects multiple aspects of the capability of the system to process queries. These aspects
include the selected database size against which the queries are executed, the query processing power when queries
Page 6
are submitted by a single stream, and the query throughput when queries are submitted by multiple concurrent
users. The TPC-H Price/Performance metric is expressed as $/QphH@Size. To be compliant with the TPC-H standard, all references to TPC-H results for a given configuration must include all required reporting components (see
Clause 6.4.6). The TPC believes that comparisons of TPC-H results measured against different database sizes are
misleading and discourages such comparisons.
The TPC-H database must be implemented using a commercially available database management system (DBMS)
and the queries executed via an interface using dynamic SQL. The specification provides for variants of SQL, as
implementers are not required to have implemented a specific SQL standard in full.
TPC-H uses terminology and metrics that are similar to other benchmarks, originated by the TPC and others. Such
similarity in terminology does not in any way imply that TPC-H results are comparable to other benchmarks. The
only benchmark results comparable to TPC-H are other TPC-H results compliant with the same revision.
Despite the fact that this benchmark offers a rich environment representative of many decision support systems, this
benchmark does not reflect the entire range of decision support requirements. In addition, the extent to which a customer can achieve the results reported by a vendor is highly dependent on how closely TPC-H approximates the
customer application. The relative performance of systems derived from this benchmark does not necessarily hold
for other workloads or environments. Extrapolations to any other environment are not recommended.
Benchmark results are highly dependent upon workload, specific application requirements, and systems design and
implementation. Relative system performance will vary as a result of these and other factors. Therefore, TPC-H
should not be used as a substitute for a specific customer application benchmarking when critical capacity planning
and/or product evaluation decisions are contemplated.
Benchmark sponsors are permitted several possible system designs, provided that they adhere to the model
described in Clause 7. A full disclosure report (FDR) of the implementation details, as specified in Clause 9, must
be made available along with the reported results.
Comment 1: While separated from the main text for readability, comments and appendices are a part of the standard and their provisions must be complied with.
Comment 2: The contents of some appendices are provided in a machine readable format and are not included in
the printed copy of this document.
0.2
General Implementation Guidelines

The rules for pricing are included in the current revision of the TPC Pricing Specification Version 1 located at
www.tpc.org.
The purpose of TPC benchmarks is to provide relevant, objective performance data to industry users. To achieve
that purpose, TPC benchmark specifications require that benchmark tests be implemented with systems, products,
technologies and pricing that:
Are generally available to users;
Are relevant to the market segment that the individual TPC benchmark models or represents (e.g., TPC-H
models and represents complex, high data volume, decision support environments);
Would plausibly be implemented by a significant number of users in the market segment the benchmark models or represents.
The use of new systems, products, technologies (hardware or software) and pricing is encouraged so long as they
meet the requirements above. Specifically prohibited are benchmark systems, products, technologies or pricing
Page 7
(hereafter referred to as "implementations") whose primary purpose is performance optimization of TPC benchmark results without any corresponding applicability to real-world applications and environments. In other words,
all "benchmark special" implementations that improve benchmark results but not real-world performance or pricing, are prohibited.
The following characteristics shall be used as a guide to judge whether a particular implementation is a benchmark
special. It is not required that each point below be met, but that the cumulative weight of the evidence be considered
to identify an unacceptable implementation. Absolute certainty or certainty beyond a reasonable doubt is not
required to make a judgment on this complex issue. The question that must be answered is: "Based on the available
evidence, does the clear preponderance (the greater share or weight) of evidence indicate that this implementation is
a benchmark special?"
The following characteristics shall be used to judge whether a particular implementation is a benchmark special:
a)
Is the implementation generally available, documented, and supported?
b)
Does the implementation have significant restrictions on its use or applicability that limits its use beyond TPC
benchmarks?
c)
Is the implementation or part of the implementation poorly integrated into the larger product?
d)
Does the implementation take special advantage of the limited nature of TPC benchmarks (e.g., query profiles,
query mix, concurrency and/or contention, isolation requirements, etc.) in a manner that would not be generally applicable to the environment the benchmark represents?
e)
Is the use of the implementation discouraged by the vendor? (This includes failing to promote the implementation in a manner similar to other products and technologies.)
f)
Does the implementation require uncommon sophistication on the part of the end-user, programmer, or system
administrator?
g)
Is the implementation (including beta) being purchased or used for applications in the market area the benchmark represents? How many sites implemented it? How many end-users benefit from it? If the implementation is not currently being purchased or used, is there any evidence to indicate that it will be purchased or used
by a significant number of end-user sites?
Comment: The characteristics listed in this clause are not intended to include the driver or implementation specific
layer, which are not necessarily commercial software, and have their own specific requirements and limitation enumerated in Clause 7. The listed characteristics and prohibitions of Clause 7 should be used to determine if the driver
or implementation specific layer is a benchmark special.
0.3
General Measurement Guidelines

TPC benchmark results are expected to be accurate representations of system performance. Therefore, there are certain guidelines that are expected to be followed when measuring those results. The approach or methodology to be
used in the measurements are either explicitly described in the specification or left to the discretion of the test sponsor.
When not described in the specification, the methodologies and approaches used must meet the following requirements:
The approach is an accepted engineering practice or standard;
The approach does not enhance the result;
Equipment used in measuring the results is calibrated according to established quality standards;
Fidelity and candor is maintained in reporting any anomalies in the results, even if not specified in the bench-
Page 8
mark requirements.
Comment: The use of new methodologies and approaches is encouraged so long as they meet the requirements
above.
Page 9
2: LOGICAL DATABASE DESIGN

2.1
Business and Application Environment

TPC Benchmark H is comprised of a set of business queries designed to exercise system functionalities in a manner representative of complex business analysis applications. These querieshave been given a realistic context, portraying the activity of a wholesale supplier to help the reader relate intuitively to the components of the benchmark.
TPC-Hdoes not represent the activity of any particular business segment, but rather any industry which must manage, sell, or distribute a product worldwide (e.g., car rental, food distribution, parts, suppliers, etc.). TPC-H does not
attempt to be a model of how to build an actual information analysis application.
The purpose of this benchmark is to reduce the diversity of operations found in an information analysis application,
while retaining the application's essential performance characteristics, namely: the level of system utilization and
the complexity of operations. A large number of queries of various types and complexities needs to be executed to
completely manage a business analysis environment. Many of the queries are not of primary interest for performance analysis because of the length of time the queries run, the system resources they use and the frequency of
their execution. The queries that have been selected exhibit the following characteristics:
They have a high degree of complexity;
They use a variety of access
They are of an ad hoc nature;
patterns; theyexamine a large percentage of the available data;
They all differ from each other;
They contain query parameters that change across query executions.
These selected queries provide answers to the following classes of business analysis:
Pricing and promotions;
Supply and demand management;
Profit and revenue management;
Customer satisfaction study;
Market share study;
Shipping management.
Although the emphasis is on information analysis, the benchmark recognizes the need to periodically refresh the
database. The database is not a one-time snapshot of a business operations database nor is it a database where OLTP
applications are running concurrently. The database must, however, be able to support queries and refresh functions
against all tables on a 7 day by 24 hour (7 x 24) basis.
While the benchmark models a business environment in which refresh functions are an integral part of data maintenance, the refresh functions actually required in the benchmark do not attempt to model this aspect of the business
environment. Their purpose is rather to demonstrate the update functionality for the DBMS, while simultaneously
assessing an appropriate performance cost to the maintenance of auxiliary data structures, such as secondary indices.
Comment: The benchmark does not include any test or measure to verify continuous database availability or particular system features which would make the benchmarked configuration appropriate for 7x24 operation. References
Page 10
to continuous availability and 7x24 operation are included in the benchmark specification to provide a more complete picture of the anticipated decision support environment. A configuration offering less that 7x24 availability
can produce compliant benchmark results as long as it meets all the requirements described in this specification.
Decision Makers
DSS Queries
DSS
Database
TPC-D
Busines
s
Business
Operations
OLTP
Database
OLTP
Transactions
Figure 1: The TPC-H Business Environment illustrates the TPC-H business environment and highlights the basic
differences between TPC-H and other TPC benchmarks.
Figure 1: The TPC-H Business Environment
Other TPC benchmarks model the operational end of the business environment where transactions are executed on
a real time basis. The TPC-H benchmark, however, models the analysis end of the business environment where
trends are computed and refined data are produced to support the making of sound business decisions. In OLTP
benchmarks the raw data flow into the OLTP database from various sources where it is maintained for some period
of time. In TPC-H, periodic refresh functions are performed against a DSS database whose content is queried on
behalf of or by various decision makers.
Page 11
2.2
Database Entities, Relationships, and Characteristics

The components of the TPC-H database are defined to consist of eight separate and individual tables (the Base
Tables). The relationships between columns of these tables are illustrated in Figure 2: The TPC-H Schema.
Figure 2: The TPC-H Schema
PART (P_)
SF*200,000
PARTSUPP (PS_)
SF*800,000
LINEITEM (L_)
SF*6,000,000
ORDERS (O_)
SF*1,500,000
PARTKEY
PARTKEY
ORDERKEY
ORDERKEY
NAME
SUPPKEY
PARTKEY
CUSTKEY
MFGR
AVAILQTY
SUPPKEY
ORDERSTATUS
BRAND
SUPPLYCOST
LINENUMBER
TOTALPRICE
TYPE
COMMENT
QUANTITY
ORDERDATE
SIZE
CONTAINER
RETAILPRICE
COMMENT
CUSTOMER (C_)
SF*150,000
CUSTKEY
NAME
ADDRESS
SUPPLIER (S_)
SF*10,000
SUPPKEY
NAME
ADDRESS
NATIONKEY
PHONE
ACCTBAL
NATIONKEY
PHONE
ACCTBAL
MKTSEGMENT
COMMENT
NATION (N_)
25
NATIONKEY
COMMENT
NAME
REGIONKEY
COMMENT
EXTENDEDPRICE
DISCOUNT
TAX
RETURNFLAG
LINESTATUS
ORDERPRIORITY
CLERK
SHIPPRIORITY
COMMENT
SHIPDATE
COMMITDATE
RECEIPTDATE
SHIPINSTRUCT
SHIPMODE
COMMENT
REGION (R_)
5
REGIONKEY
NAME
COMMENT
Legend:
The parentheses following each table name contain the prefix of the column names for that table;
The arrows point in the direction of the one-to-many relationships between tables;
The number/formula below each table name represents the cardinality (number of rows) of the table. Some are
factored by SF, the Scale Factor, to obtain the chosen database size. The cardinality for the LINEITEM table is
approximate (see Clause 5.2.5).
Page 12
2.3
Datatype Definitions
2.3.1
The following datatype definitions apply to the list of columns of each table:
Identifier means that the column must be able to hold any key value generated for that column and be able to
support at least 2,147,483,647 unique values;
Comment: A common implementation of this datatype will be an integer. However, for SF greater than 300 some
column values will exceed the range of integer values supported by a 4-byte integer. A test sponsor may use some
other datatype such as 8-byte integer, decimal or character string to implement the identifier datatype;
Integer means that the column must be able to exactly represent integer values (i.e., values in increments of 1)
in the range of at least -2,147,483,646 to 2,147,483,647.
Decimal means that the column must be able to represent values in the range -9,999,999,999.99 to
+9,999,999,999.99 in increments of 0.01; the values can be either represented exactly or interpreted to be in
this range;
Big Decimal is of the Decimal datatype as defined above, with the additional property that it must be large
enough to represent the aggregated values stored in temporary tables created within query variants;
Fixed text, size N means that the column must be able to hold any string of characters of a fixed length of N.
Comment: If the string it holds is shorter than N characters, then trailing spaces must be stored in the database or
the database must automatically pad with spaces upon retrieval such that a CHAR_LENGTH() function will return
N.
Variable text, size N means that the column must be able to hold any string of characters of a variable length
with a maximum length of N. Columns defined as "variable text, size N" may optionally be implemented as
"fixed text, size N";
Date is a value whose external representation can be expressed as YYYY-MM-DD, where all characters are
numeric. A date must be able to express any day within at least 14 consecutive years. There is no requirement
specific to the internal representation of a date.
Comment: The implementation datatype chosen by the test sponsor for a particular datatype definition must be
applied consistently to all the instances of that datatype definition in the schema, except for identifier columns,
whose datatype may be selected to satisfy database scaling requirements.
2.3.2
The symbol SF is used in this document to represent the scale factor for the database (see Clause 5).
2.4
Table Layouts
2.4.1
Required Tables
The following list defines the required structure (list of columns) of each table. The annotations for primary keys
and foreign references are for clarification only and do not specify any implementation requirement such as integrity constraints:
PART Table Layout
Column Name
Datatype Requirements
Comment
P_PARTKEY
identifier
SF*200,000 are populated
P_NAME
variable text, size 55
P_MFGR
fixed text, size 25
Page 13
P_BRAND
fixed text, size 10
P_TYPE
P_SIZE
integer
P_CONTAINER
fixed text, size 10
P_RETAILPRICE
decimal
P_COMMENT
Primary Key: P_PARTKEY

SUPPLIER Table Layout
Column Name
Comment
S_SUPPKEY
identifier
S_NAME
fixed text, size 25
S_ADDRESS
S_NATIONKEY
identifier
S_PHONE
fixed text, size 15
S_ACCTBAL
decimal
S_COMMENT
Foreign key reference to N_NATIONKEY
Primary Key: S_SUPPKEY

PARTSUPP Table Layout
Column Name
Comment
PS_PARTKEY
identifier
Foreign key reference to P_PARTKEY
PS_SUPPKEY
identifier
Foreign key reference to S_SUPPKEY
PS_AVAILQTY
integer
PS_SUPPLYCOST
decimal
PS_COMMENT
Compound Primary Key: PS_PARTKEY, PS_SUPPKEY

CUSTOMER Table Layout
Column Name
Comment
C_CUSTKEY
identifier
C_NAME
C_ADDRESS
C_NATIONKEY
identifier
C_PHONE
fixed text, size 15
C_ACCTBAL
decimal
C_MKTSEGMENT
fixed text, size 10
C_COMMENT
Foreign key reference to N_NATIONKEY
Primary Key: C_CUSTKEY

Page 14
ORDERS Table Layout

Column Name
Comment
O_ORDERKEY
identifier
SF*1,500,000 are sparsely populated
O_CUSTKEY
identifier
Foreign key reference to C_CUSTKEY
O_ORDERSTATUS
fixed text, size 1
O_TOTALPRICE
decimal
O_ORDERDATE
date
O_ORDERPRIORITY
fixed text, size 15
O_CLERK
fixed text, size 15
O_SHIPPRIORITY
integer
O_COMMENT
Primary Key: O_ORDERKEY

Comment: Orders are not present for all customers. In fact, one-third of the customers do not have any order in the
database. The orders are assigned at random to two-thirds of the customers (see Clause 5). The purpose of this is to
exercise the capabilities of the DBMS to handle "dead data" when joining two or more tables.
LINEITEM Table Layout
Column Name
Comment
L_ORDERKEY
identifier
Foreign key reference to O_ORDERKEY
L_PARTKEY
identifier
Foreign key reference to P_PARTKEY, Compound Foreign Key Reference to (PS_PARTKEY, PS_SUPPKEY) with L_SUPPKEY
L_SUPPKEY
identifier
Foreign key reference to S_SUPPKEY, Compound Foreign key reference to (PS_PARTKEY, PS_SUPPKEY) with L_PARTKEY
L_LINENUMBER
integer
L_QUANTITY
decimal
L_EXTENDEDPRICE
decimal
L_DISCOUNT
decimal
L_TAX
decimal
L_RETURNFLAG
fixed text, size 1
L_LINESTATUS
fixed text, size 1
L_SHIPDATE
date
L_COMMITDATE
date
L_RECEIPTDATE
date
L_SHIPINSTRUCT
fixed text, size 25
L_SHIPMODE
fixed text, size 10
L_COMMENT
variable text size 44
Compound Primary Key: L_ORDERKEY, L_LINENUMBER
Page 15
NATION Table Layout

Column Name
Comment
N_NATIONKEY
identifier
25 nations are populated
N_NAME
fixed text, size 25
N_REGIONKEY
identifier
N_COMMENT
Foreign key reference to R_REGIONKEY
Primary Key: N_NATIONKEY

REGION Table Layout
Column Name
Comment
R_REGIONKEY
identifier
5 regions are populated
R_NAME
fixed text, size 25
R_COMMENT
Primary Key: R_REGIONKEY

2.4.2
Constraints
The use of constraints is optional. There is no specific requirement to define primary keys, foreign keys or check
constraints. However, if constraints are used, they must satisfy the following requirements:
They must be specified using SQL. There is no specific implementation requirement. For example, CREATE
TABLE, ALTER TABLE, and CREATE TRIGGER are all valid statements;
Constraints must be enforced either at the statement level or at the transaction level;
All defined constraints must be enforced and validated before the load test is complete (see Clause 6.1.1.2);
Any subset of the constraints listed below may be specified. No additional constraints may be used.
2.4.2.1
Nulls: The NOT NULL attribute may be used for any column.
2.4.2.2
Primary keys: The following primary keys may be defined as primary key (using the PRIMARY KEY clause or
other equivalent syntax):
P_PARTKEY;
S_SUPPKEY;
PS_PARTKEY, PS_SUPPKEY;
C_CUSTKEY;
O_ORDERKEY;
L_ORDERKEY, L_LINENUMBER;
N_NATIONKEY;
R_REGIONKEY.
Constraining a column (or set of columns) to contain unique values can only be implemented for the primary key(s)
listed above.
Page 16
2.4.2.3
Foreign Keys: Any of the foreign keys listed in the comments of Clause 2.4.1 may be defined. There is no specific
requirement for delete/update actions (e.g., RESTRICT, CASCADE, NO ACTION, etc.). If any foreign key relationship is defined by an implementation, then all foreign key relationships must be defined by the implementation.
2.4.2.4
Check Constraints: Check constraints may be defined to restrict the database contents. In order to support evolutionary change, the check constraints must not rely on knowledge of the enumerated domains of each column. The
following list of expressions defines permissible check constraints:
1. Positive Keys
1.P_PARTKEY >= 0
2.S_SUPPKEY >= 0
3.C_CUSTKEY >= 0
4.PS_PARTKEY >= 0
5.R_REGIONKEY >= 0
6.N_NATIONKEY >= 0
2. Open-interval constraints
1.P_SIZE >= 0
2.P_RETAILPRICE >= 0
3.PS_AVAILQTY >= 0
4.PS_SUPPLYCOST >= 0
5.O_TOTALPRICE >= 0
6.L_QUANTITY >= 0
7.L_EXTENDEDPRICE >= 0
8.L_TAX >= 0
3. Closed-interval constraints
1.L_DISCOUNT between 0.00 and 1.00
4. Multi-column constraints
1.L_SHIPDATE <= L_RECEIPTDATE
Comment: The constraints rely solely on the diagram provided in Clause 2.2 and the description in Clause 2.4.
They are not derived from explicit knowledge of the data population specified in Clause 5.2.
2.5
Implementation Rules
2.5.1
The database shall be implemented using a commercially available database management system (DBMS).
2.5.2
The physical clustering of records within the database is allowed as long as this clustering does not alter the logical
independence of each table.
Comment: The intent of this clause is to permit flexibility in the physical design of a database while preserving a
strict logical view of all the tables.
Page 17
2.5.3
At the end of the Load Test, all tables must have exactly the number of rows defined for the scale factor, SF, and the
database population, both specified in Clause 5.
2.5.4
Horizontal partitioning of base tables or auxiliary structures created by database directives (see Clause 2.5.7) is
allowed. Groups of rows from a table or auxiliary structure may be assigned to different files, disks, or areas. If this
assignment is a function of data in the table or auxiliary structure, the assignment must be based on the value of a
partitioning field. A partitioning field must be one and only one of the following:
A primary key as defined in Clause 1.4.2.2
A foreign key as defined in Clause 1.4.1 may be defined. There is no specific requirement for delete/update
actions (e.g., RESTRICT, CASCADE, NO ACTION, etc.). If any foreign key relationship is defined by an
implementation, then all foreign key relationships must be defined by the implementation.
A single date column
Some partitioning schemes require the use of directives that specify explicit values for the partitioning field. If such
directives are used they must satisfy the following conditions:
They may not rely on any knowledge of the data stored in the table except the minimum and maximum values
of columns used for the partitioning field. The minimum and maximum values of columns are specified in
Clause 5.2.3
Within the limitations of integer division, they must define each partition to accept an equal portion of the
range between the minimum and maximum values of the partitioning column(s).
The directives must allow the insertion of values of the partitioning column(s) outside the range covered by
the minimum and maximum values, as required by The database must allow for insertion of arbitrary data values that conform to the datatype and optional constraint definitions from Clause 2.3 and Clause 2.4..
Multiple-level partitioning of base tables or auxiliary structures is allowed only if each level of partitioning satisfies
the conditions stated above and each level references only one partitioning field as defined above. If implemented,
the details of such partitioning must be disclosed.
2.5.5
Physical placement of data on durable media is not auditable. SQL DDL that explicitly partitions data vertically is
prohibited. The row must be logically presented as an atomic set of columns.
Comment: This implies that vertical partitioning which does not rely upon explicit partitioning directives is
allowed. Explicit partitioning directives are those that assign groups of columns of one row to files, disks or areas
different from those storing the other columns in that row.
2.5.6
2.5.7
Except as provided in Clause 2.5.7, logical replication of database objects (i.e., tables, rows, or columns) is not
allowed. The physical implementation of auxiliary data structures to the tables may involve data replication of
selected data from the tables provided that:
All replicated data are managed by the DBMS, the operating system, or the hardware;
All replications are transparent to all data manipulation operations;
Data modifications are reflected in all logical copies of the replicated data by the time the updating transaction is committed;
All copies of replicated data maintain full ACID properties (see Clause 4) at all times.
Auxiliary data structures that constitute logical replications of data from one or more columns of a base table (e.g.,
indexes, materialized views, summary tables, structures used to enforce relational integrity constraints) must conform to the provisions of Clause 2.5.6. The directives defining and creating these structures are subject to the folTPC Benchmark H Standard Specification Revision 2.8.0
Page 18
lowing limitations:
They may reference no more than one base table, and may not reference other auxiliary structures.
They must satisfy exactly one of the following two conditions:
They may reference no more than one base table column that is chosen from:
A column that is a primary key on its own, or is a component of a compound primary key as
defined in Required Tables
A column that is a foreign key on its own, or is a component of a compund foreign key as defined
in Required Tables.
A column having a date datatype as defined in Clause 2.3.
They may reference more than one column of a base table only if those columns exactly comprise a compound primary or foreign key of that table, as defined in Clause 2.4.1.
They may contain functions or expressions on explicitly permitted columns
No directives (e.g. DDL, session options, global configuration parameters) are permitted in TPC-HTPC-H scripts
whose effect is to cause the materialization of columns (or functions on columns) in auxiliary data structures other
than those columns explicitly permitted by the above limitations. Further, no directives are permitted whose effect
is to cause the materialization of columns in auxiliary data structures derived from more than one table.
Comment: Database implementations of auxiliary structures generated as a result of compliant directives usually
contain embedded pointers or references to corresponding base table rows. Database implementations that transparently employ either "row IDs" or embedded base table primary key values for this purpose are equally acceptable. In particular, the generation of transparently embedded primary key values required by auxiliary structures is a
permitted materialization of the primary key column(s). Primary and foreign key columns are defined in Required
Tables.
2.5.8
Table names should match those provided in Clause 2.4. In cases where a table name conflicts with a reserved word
in a given implementation, delimited identifiers or an alternate meaningful name may be chosen.
2.5.9
For each table, the set of columns must include all those defined in Clause 2.4. No column can be added to any of
the tables. However, the order of the columns is not constrained.
2.5.10
Column names must match those provided in Clause 2.4.
2.5.11
Each column, as described in Clause 2.4, must be logically discrete and independently accessible by the data manager. For example, C_ADDRESS and C_PHONE cannot be implemented as two sub-parts of a single discrete column C_DATA.
2.5.12
Each column, as described in Clause 2.4, must be accessible by the data manager as a single column. For example,
P_TYPE cannot be implemented as two discrete columns P_TYPE1 and P_TYPE2.
2.5.13
The database must allow for insertion of arbitrary data values that conform to the datatype and optional constraint
definitions from Clause 2.3 and Clause 2.4.
Comment 1: Although the refresh functions (see Clause 3.26) do not insert arbitrary values and do not modify all
tables, all tables must be modifiable throughout the performance test.
Page 19
Comment 2: The intent of this Clause is to prevent the database schema definition from taking undue advantage of
the limited data population of the database (see also Clause 1.2 and Clause 6.2.7).
2.6
Data Access Transparency Requirements
2.6.1
Data Access Transparency is the property of the system that removes from the query text any knowledge of the
location and access mechanisms of partitioned data. No finite series of tests can prove that the system supports
complete data access transparency. The requirements below describe the minimum capabilities needed to establish
that the system provides transparent data access. An implementation that uses horizontal partitioning must meet the
requirements for transparent data access described in Clause 2.6.2 and Clause 2.6.3.
Comment: The intent of this Clause is to require that access to physically and/or logically partitioned data be provided directly and transparently by services implemented by commercially available layers such as the interactive
SQL interface, the database management system (DBMS), the operating system (OS), the hardware, or any combination of these.
2.6.2
Each of the tables described in Clause 2.4 must be identifiable by names that have no relationship to the partitioning
of tables. All data manipulation operations in the executable query text (see Clause 3.1.1.2) must use only these
names.
2.6.3
Using the names which satisfy Clause 2.6.2, any arbitrary non-TPC-H query must be able to reference any set of
rows or columns:
Identifiable by any arbitrary condition supported by the underlying DBMS;
Using the names described in Clause 2.6.2 and using the same data manipulation semantics and syntax for all
tables.
For example, the semantics and syntax used to query an arbitrary set of rows in any one table must also be usable
when querying another arbitrary set of rows in any other table.
Comment: The intent of this clause is that each TPC-H query uses general purpose mechanisms to access data in
the database.
Page 20
3: QUERIES AND REFRESH FUNCTIONS

This Clause describes the twenty-two decision support queries and the two database refresh functions that must be
executed as part of the TPC-H benchmark.
3.1
General Requirements and Definitions for Queries
3.1.1
Query Overview
3.1.1.1
Each query is defined by the following components:
The business question, which illustrates the business context in which the query could be used;
The functional query definition, which defines, using the SQL-92 language, the function to be performed by
the query;
The substitution parameters, which describe how to generate the values needed to complete the query syntax;
The query validation, which describes how to validate the query against the qualification database.
3.1.1.2
For each query, the test sponsor must create an implementation of the functional query definition, referred to as the
executable query text.
3.1.2
Functional Query Definitions
3.1.2.1
The functional query definitions are written in the SQL-92 language (ISO/IEC 9075:1992), annotated where necessary to specify the number of rows to be returned. They define the function that each executable query text must
perform against the test database (see Clause 5.1.1).
3.1.2.2
If an executable query text, with the exception of its substitution parameters, is not identical to the specified functional query definition it must satisfy the compliance requirements of Clause 3.2.
3.1.2.3
When a functional query definition includes the creation of a new entity (e.g., cursor, view, or table) some mechanism must be used to ensure that newly created entities do not interfere with other execution streams and are not
shared between multiple execution streams (see Clause 6.1.2.3).
Functional query definitions in this document (as well as QGEN, see Clause 3.1.4) achieve this separation by
appending a text-token to the new entity name. This text-token is expressed in upper case letters and enclosed in
square brackets (i.e., [STREAM_ID]). This text-token, whenever found in the functional query definition, must be
replaced by a unique stream identification number (starting with 0) to complete the executable query text.
Comment: Once an identification number has been generated and assigned to a given query stream, the same identification number must be used for that query stream for the duration of the test.
3.1.2.4
When a functional query definition includes the creation of a table, the datatype specification of the columns uses
the <datatype> notation. The definition of <datatype> is obtained from Clause 2.3.1.
3.1.2.5
Any entity created within the scope of an executable query text must also be deleted within the scope of that same
executable query text.
3.1.2.6
A logical tablespace is a named collection of physical storage devices referenced as a single, logically contiguous,
non-divisible entity.
Page 21
3.1.2.7
If CREATE TABLE statements are used during the execution of the queries, these CREATE TABLE statements
may be extended only with a tablespace reference (e.g., IN <tablespacename>). A single tablespace must be used
for all these tables.
Comment: The allowance for tablespace syntax applies only to variants containing CREATE TABLE statements.
3.1.2.8
All tables created during the execution of a query must meet the ACID properties defined in Clause 4.
3.1.2.9
Queries 2, 3, 10, 18 and 21 require that a given number of rows are to be returned (e.g., Return the first 10 selected
rows). If N is the number of rows to be returned, the query must return exactly the first N rows unless fewer than N
rows qualify, in which case all rows must be returned. There are three permissible ways of satisfying this requirement. A test sponsor must select any one of them and use it consistently for all the queries that require that a specified number of rows be returned.
1. Vendor-specific control statements supported by a test sponsors interactive SQL interface may be used (e.g.,
SET ROWCOUNT n) to limit the number of rows returned.
2. Control statements recognized by the implementation specific layer (see Clause 7.2.4) and used to control a loop
which fetches the rows may be used to limit the number of rows returned (e.g., while rowcount <= n).
3. Vendor-specific SQL syntax may be added to the SELECT statement to limit the number of rows returned (e.g.,
SELECT FIRST n). This syntax is not classified as a minor query modification since it completes the functional
requirements of the functional query definition and there is no standardized syntax defined. In all other respects,
the query must satisfy the requirements of Clause 3.2. The syntax must deal solely with the answer set, and must
not make any additional explicit reference, for example to tables, indices, or access paths.
3.1.3
Substitution Parameters and Output Data
3.1.3.1
Each query has one or more substitution parameters. When generating executable query text a value must be supplied for each substitution parameter of that query. These values must be used to complete the executable query
text. These substitution parameters are expressed as names in uppercase and enclosed in square brackets. For example, in the Pricing Summary Report Query (see Clause 3.4) the substitution parameter [DELTA], whenever found in
the functional query definition, must be replaced by the value generated for DELTA to complete the executable
query text.
Comment 1: When dates are part of the substitution parameters, they must be expressed in a format that includes
the year, month and day in integer form, in that order (e.g., YYYY-MM-DD). The delimiter between the year,
month and day is not specified. Other date representations, for example the number of days since 1970-01-01, are
specifically not allowed.
Comment 2: When a substitution parameter appears more than once in a query, a single value is generated for that
substitution parameter and each of its occurrences in the query must be replaced by that same value.
Comment 3: Generating executable query text may also involve additional text substitution (see Clause 3.1.2.3).
3.1.3.2
The term randomly selected when used in the definitions of substitution parameters means selected at random
from a uniform distribution over the range or list of values specified.
3.1.3.3
Seeds to the random number generator used to generate substitution parameters must be selected using the following method:
An initial seed (seed0) is first selected as the time stamp of the end of the database load time expressed in the format
mmddhhmmss where mm is the month, dd the day, hh the hour, mm the minutes and ss the seconds. This seed is
used to seed the Power test of Run 1. Further seeds (for the Throughput test) are chosen as seed0 + 1, seed0 +
Page 22
2,...,seed0 + n where s is the number of throughput streams selected by the vendor. This process leads to s + 1 seeds
required for Run 1 of a benchmark with s streams. The seeds for Run 2 can be the same as those for Run 1 (see
5.3.2). However, should the test sponsor decide to use different seeds for Run 2 from those used for Run 1, the
sponsor must use a selection process similar to that of Run 1. The seeds must again be of the form seed0, seed0 + 1,
seed0 + 2,...., seed0 + s, where and seed0 is be the time stamp of the end of Run 1, expressed in the format defined
above.
Comment 1: The intent of this Clause is to prevent performance advantage that could result from multiple streams
beginning work with identical seeds or using seeds known in advance while providing a well-defined and unified
method for seed selection.
Comment 2: QGEN is a utility provided by the TPC (see Clause 3.1.4) to generate executable query text. If a sponsor-created tool is used instead of QGEN, the behavior of its seeds must satisfy this Clause and its code must be disclosed. After execution, the query returns one or more rows. The rows returned are either rows from the database or
rows built from data in the database and are called the output data.
3.1.3.4
3.1.3.5
Output data for each query should be expressed in a format easily readable by a non-sophisticated computer user. In
particular, in order to be comparable with known output data for the purpose of query validation (see Clause 3.3),
the format of the output data for each query must adhere to the following guidelines:
a)
Columns appear in the order specified by the SELECT list of either the functional query definition or an
approved variant. Column headings are optional.
b)
Non-integer expressions including prices are expressed in decimal notation with at least two digits behind the
decimal point.
c)
Integer quantities contain no leading zeros.
d)
Dates are expressed in a format that includes the year, month and day in integer form, in that order (e.g.,
YYYY-MM-DD). The delimiter between the year, month and day is not specified. Other date representations,
for example the number of days since 1970-01-01, are specifically not allowed.
e)
Strings are case-sensitive and must be displayed as such. Leading or trailing blanks are acceptable.
f)
The amount of white space between columns is not specified.
The precision of all values contained in the query validation output data must adhere to the following rules:
a)
For singleton column values and results from COUNT aggregates, the values must exactly match the query
validation output data.
b)
For ratios, results must be within 1% of the query validation output data when reported to the nearest 1/100th,
rounded up.
c)
For results from SUM aggregates, the resulting values must be within $100 of the query validation output
data.
d)
For results from AVG aggregates, the resulting values must be within 1% of the query validation output data
when reported to the nearest 1/100th, rounded up.
3.1.4
The QGEN Program
3.1.4.1
Executable query text must be generated according to the requirements of Clause 3.1.2 and Clause 3.1.3. The
QGEN source code provided in Appendix D is a sample implementation of an executable query text generator. It
has been written in ANSI 'C' and has been ported to a large number of platforms. If QGEN is used, its version and
the release numbers must match the version and the release numbers of the benchmark specification.
Page 23
Comment 1: Use of QGEN is strongly recommended. Exact query answer set compliance is required. This may not
be possible unless substitution parameters and text tokens are generated and integrated within the executable query
text identically to QGEN's output.
Comment 2: The numbering used in this Clause for the definition of substitution parameters corresponds to the
numbering used by QGEN to generate values for these substitution parameters.
3.2
Query Compliance
3.2.1
The queries must be expressed in a commercially available implementation of the SQL language. Since the latest
ISO SQL standard (currently ISO/IEC 9075:1992) has not yet been fully implemented by most vendors, and since
the ISO SQL language is continually evolving, the TPC-H benchmark specification includes a number of permissible deviations from the formal functional query definitions found in Clause 3. An on-going process is also defined
to approve additional deviations that meet specific criteria.
3.2.2
There are two types of permissible deviations from the functional query definitions, as follows:
a)
Minor query modifications;
b)
Approved query variants.
3.2.3
Minor Query Modifications
3.2.3.1
It is recognized that implementations require specific adjustments for their operating environment and the syntactic
variations of its dialect of the SQL language. Therefore, minor query modifications are allowed. Minor query modifications are those that fall within the bounds of what is described in Clause 3.2.3.3. They do not require approval.
Modifications that do not fall within the bounds of what is described in Clause 3.2.3.3 are not minor and are not
compliant unless they are an integral part of an approved query variant (see Clause 3.2.4).
Comment 1: The intent of this Clause is to allow the use of any number of minor query modifications. These query
modifications are labeled minor based on the assumption that they do not significantly impact the performance of
the queries.
Comment 2: The only exception is for the queries that require a given number of rows to be returned. The requirements governing this exception are given in Clause 3.1.2.9.
3.2.3.2
Minor query modifications can be used to produce executable query text by modifying either a functional query
definition or an approved variant of that definition.
3.2.3.3
The following query modifications are minor:

a)
Table names - The table and view names found in the CREATE TABLE, CREATE VIEW, DROP VIEW and
in the FROM clause of each query may be modified to reflect the customary naming conventions of the system under test.
b)
Select-list expression aliases - For queries that include the definition of an alias for a SELECT-list item (e.g.,
AS CLAUSE), vendor-specific syntax may be used instead of the specified SQL-92 syntax. Replacement syntax must have equivalent semantic behavior. Examples of acceptable implementations include "TITLE
<string>", or "WITH HEADING <string>". Use of a select-list expression alias is optional.
c)
Date expressions - For queries that include an expression involving manipulation of dates (e.g., adding/subtracting days/months/years, or extracting years from dates), vendor-specific syntax may be used instead of the
specified SQL-92 syntax. Replacement syntax must have equivalent semantic behavior. Examples of acceptable implementations include "YEAR(<column>)" to extract the year from a date column or "DATE(<date>)
+ 3 MONTHS" to add 3 months to a date.
Page 24
d)
GROUP BY and ORDER BY - For queries that utilize a view, nested table-expression, or select-list alias
solely for the purposes of grouping or ordering on an expression, vendors may replace the view, nested tableexpression or select-list alias with a vendor-specific SQL extension to the GROUP BY or ORDER BY clause.
Examples of acceptable implementations include "GROUP BY <ordinal>", "GROUP BY <expression>",
"ORDER BY <ordinal>", and "ORDER BY <expression>".
e)
Command delimiters - Additional syntax may be inserted at the end of the executable query text for the purpose of signaling the end of the query and requesting its execution. Examples of such command delimiters are
a semicolon or the word "GO".
f)
Output formatting functions - Scalar functions whose sole purpose is to affect output formatting or intermediate arithmetic result precision (such as CASTs) may be applied to items in the outermost SELECT list of the
query.
g)
Transaction control statements - A CREATE/DROP TABLE or CREATE/DROP VIEW statement may be followed by a COMMIT WORK statement or an equivalent vendor-specific transaction control statement.
h)
Correlation names Table-name aliases may be added to the executable query text. The keyword "AS" before
the table-name alias may be omitted.
i)
Explicit ASC - ASC may be explicitly appended to columns in the ORDER BY.
j)
CREATE TABLE statements may be augmented with a tablespace reference conforming to the requirements
of Clause 3.1.2.6.
k)
In cases where identifier names conflict with SQL-92 reserved words in a given implementation, delimited
identifiers may be used.
l)
Relational operators - Relational operators used in queries such as "<", ">", "<>", "<=", and "=", may be
replaced by equivalent vendor-specific operators, for example ".LT.", ".GT.", "!=" or "^=", ".LE.", and "==",
respectively.
m)
Nested table-expression aliasing - For queries involving nested table-expressions, the nested keyword "AS"
before the table alias may be omitted.
n)
If an implementation is using variants involving views and the implementation only supports DROP
RESTRICT semantics (i.e., all dependent objects must be dropped first), then additional DROP statements
for the dependent views may be added.
o)
At large scale factors, the aggregates may exceed the range of the values supported by an integer. The aggregate functions AVG and COUNT may be replaced with equivalent vendor-specific functions to handle the
expanded range of values (e.g., AVG_BIG and COUNT_BIG).
p)
Substring Scalar Functions For queries which use the SUBSTRING() scalar function, vendor-specific syntax
may be used instead of the specified SQL 92 syntax. Replacement syntax must have equivalent semantic
behavior. For example, SUBSTRING(C_PHONE, 1, 2).
q)
Outer Join For outer join queries, vendor specific syntax may be used instead of the specified SQL 92 syntax. Replacement syntax must have equivalent semantic behavior. For example, the join expression CUSTOMER LEFT OUTER JOIN ORDERS ON C_CUSTKEY = O_CUSTKEY may be replaced by adding
CUSTOMER and ORDERS to the from clause and adding a specially-marked join predicate (e.g.,
C_CUSTKEY *= O_CUSTKEY).
3.2.3.4
The application of minor query modifications to functional query definitions or approved variants must be consistent over the query set. For example, if a particular vendor-specific date expression or table name syntax is used in
one query, it must be used in all other queries involving date expressions or table names.
3.2.3.5
The use of minor modifications to obtain executable query text must be disclosed and justified (see Clause 9.3.3.4).
Page 25
3.2.4
Approved Query Variants
3.2.4.1
Approval of any new query variant is required prior to using such variant to produce compliant TPC-H results. The
approval process is based on criteria defined in Clause 3.2.4.3.
3.2.4.2
Query variants that have already been approved are listed in Appendix B of this specification.
Comment: Since Appendix B is updated each time a new variant is approved, test sponsors should obtain the latest
version of this appendix prior to implementing the benchmark.
3.2.4.3
3.2.4.4
The executable query text for each query in a compliant implementation must be taken from either the functional
query definition (see Clause 3) or an approved query variant (see Appendix B). Except as specifically allowed in
Clause 3.2.3.3, executable query text must be used in full exactly as written in the TPC-H specification. New query
variants will be considered for approval if they meet one of the following criteria:
a)
The vendor cannot successfully run the executable query text against the qualification database using the functional query definition or an approved variant even after applying appropriate minor query modifications as
per Clause 3.2.3.
b)
The variant contains new or enhanced SQL syntax, relevant to the benchmark, which is defined in an
Approved Committee Draft of a new ISO SQL standard.
c)
The variant contains syntax that brings the proposed variant closer to adherence to an ISO SQL standard.
d)
The variant contains minor syntax differences that have a straightforward mapping to ISO SQL syntax used in
the functional query definition and offers functionality substantially similar to the ISO SQL standard.
To be approved, a proposed variant should have the following properties. Not all of the following properties are
specifically required. Rather, the cumulative weight of each property satisfied by the proposed variant will be the
determining factor in approving it.
a)
Variant is syntactical only, seeking functional compatibility and not performance gain.
b)
Variant is minimal and restricted to correcting a missing functionality.
c)
Variant is based on knowledge of the business question rather than on knowledge of the system under test
(SUT) or knowledge of specific data values in the test database.
d)
Variant has broad applicability among different vendors.
e)
Variant is non procedural.
f)
Variant is an SQL-92 standard [ISO/IEC 9075:1992] implementation of the functional query definition.
g)
Variant is sponsored by a vendor who can implement it and who intends on using it in an upcoming implementation of the benchmark.
3.2.4.5
Query variants that are submitted for approval will be recorded, along with a rationale describing why they were or
were not approved.
3.2.4.6
Query variants listed in Appendix B are defined using the conventions defined for functional query definitions (see
Clause 3.1.2.3 through Clause 3.1.2.6).
3.2.5
Coding Style
Implementers may code the executable query text in any desired coding style, including:
a) additional line breaks, tabs or white space
Page 26
b)
choice of upper or lower case text
The coding style used must have no impact on the performance of the system under test, and must be consistently
applied across the entire query set. Any coding style that differs from the functional query definitions in Clause 3
must be disclosed.
Comment: This does not preclude the auditor from verifying that the coding style does not affect performance.
3.3
Query Validation
3.3.1
To validate the compliance of the executable query text, the following validation test must be executed by the test
sponsor and the results reported in the full disclosure report:
1. A qualification database must be built in a manner substantially the same as the test database (see Clause 5.1.2).
2. The query validation test must be run using a qualification database that has not been modified by any update
activity (e.g., RF1, RF2, or ACID Transaction executions).
3. The query text used (see Clause 3.1.3) must be the same as that used in the performance test. The default substitution parameters provided for each query must be used. The refresh functions, RF1 and RF2, are not executed.
4. The same driver and implementation specific layer used to execute the queries against the test database must be
used for the validation of the qualification database.
5. The resulting output must match the output data specified for the query validation (see Appendix C). A subset of
this output can be found as part of the definition of each query.
6. Any difference between the output obtained and the query validation output must satisfy the requirements of
Clause 3.1.3.5.
Any query whose output differs from the query validation output to a greater degree than allowed by Clause 3.1.3.5
when run against the qualification database as specified above is not compliant.
Comment: The validation test, above, provides a minimum level of assurance of compliance. The auditor may
request additional assurance that the query texts execute in accordance with the benchmark requirements.
3.3.2
No aspect of the System Under Test (e.g., system parameters and conditional software features such as those listed
in Clause 6.2.7, hardware configuration, software releases, etc.), may differ between this demonstration of compliance and the performance test.
Comment: While the intent of this validation test is that it be executed without any change to the hardware configuration, building the qualification database on additional disks (i.e., disks not included in the priced system) is
allowed as long as this change has no impact on the results of the demonstration of compliance.
Page 27
3.4
Pricing Summary Report Query (Q1)

This query reports the amount of business that was billed, shipped, and returned.
3.4.1
Business Question
The Pricing Summary Report Query provides a summary pricing report for all lineitems shipped as of a given date.
The date is within 60 - 120 days of the greatest ship date contained in the database. The query lists totals for
extended price, discounted extended price, discounted extended price plus tax, average quantity, average extended
price, and average discount. These aggregates are grouped by RETURNFLAG and LINESTATUS, and listed in
ascending order of RETURNFLAG and LINESTATUS. A count of the number of lineitems in each group is
included.
3.4.2
Functional Query Definition
select
l_returnflag,
l_linestatus,
sum(l_quantity) as sum_qty,
sum(l_extendedprice) as sum_base_price,
sum(l_extendedprice*(1-l_discount)) as sum_disc_price,
sum(l_extendedprice*(1-l_discount)*(1+l_tax)) as sum_charge,
avg(l_quantity) as avg_qty,
avg(l_extendedprice) as avg_price,
avg(l_discount) as avg_disc,
count(*) as count_order
from
lineitem
where
l_shipdate <= date '1998-12-01' - interval '[DELTA]' day (3)
group by
l_returnflag,
l_linestatus
order by
l_returnflag,
l_linestatus;
3.4.3
Substitution Parameters
Values for the following substitution parameter must be generated and used to build the executable query text:
7. DELTA is randomly selected within [60. 120].
Comment: 1998-12-01 is the highest possible ship date as defined in the database population. (This is ENDDATE 30). The query will include all lineitems shipped before this date minus DELTA days. The intent is to choose
DELTA so that between 95% and 97% of the rows in the table are scanned.
3.4.4
Query Validation
For validation against the qualification database the query must be executed using the following values for substitution parameters and must produce the following output data:
Page 28
Values for substitution parameters:

1. DELTA = 90.
Query validation output data:
L_RETURNFLAG
L_LINESTATUS
SUM_QTY
SUM_BASE_PRICE
SUM_DISC_PRICE
37734107.00
56586554400.73
53758257134.87
991417.00
1487504710.38
1413082168.05
74476040.00
111701729697.74
106118230307.61
37719753.00
56568041380.90
53741292684.60
SUM_CHARGE
AVG_QTY
AVG_PRICE
AVG_DISC
COUNT_ORDER
55909065222.83
25.52
38273.13
.05
1478493
1469649223.19
25.52
38284.47
.05
38854
110367043872.50
25.50
38249.12
.05
2920374
55889619119.83
25.51
38250.86
.05
1478870
Page 29
3.5
Minimum Cost Supplier Query (Q2)

This query finds which supplier should be selected to place an order for a given part in a given region.
3.5.1
Business Question
The Minimum Cost Supplier Query finds, in a given region, for each part of a certain type and size, the supplier
who can supply it at minimum cost. If several suppliers in that region offer the desired part type and size at the same
(minimum) cost, the query lists the parts from suppliers with the 100 highest account balances. For each supplier,
the query lists the supplier's account balance, name and nation; the part's number and manufacturer; the supplier's
address, phone number and comment information.
3.5.2
Return the first 100 selected rows

select
s_acctbal,
s_name,
n_name,
p_partkey,
p_mfgr,
s_address,
s_phone,
s_comment
from
part,
supplier,
partsupp,
nation,
region
where
p_partkey = ps_partkey
and s_suppkey = ps_suppkey
and p_size = [SIZE]
and p_type like '%[TYPE]'
and s_nationkey = n_nationkey
and n_regionkey = r_regionkey
and r_name = '[REGION]'
and ps_supplycost = (
select
min(ps_supplycost)
from
partsupp, supplier,
nation, region
where
and s_suppkey = ps_suppkey
Page 30
)
order by
s_acctbal desc,
n_name,
s_name,
p_partkey;
3.5.3
1. SIZE is randomly selected within [1. 50];
2. TYPE is randomly selected within the list Syllable 3 defined for Types in Clause 5.2.2.13;
3. REGION is randomly selected within the list of values defined for R_NAME in Clause 5.2.3.
3.5.4
Query Validation
1. SIZE = 15;
2. TYPE = BRASS;
3. REGION = EUROPE.
S_ACCTBAL
S_NAME
N_NAME
P_PARTKEY
P_MFGR
9938.53
Supplier#000005359
UNITED KINGDOM
185358
Manufacturer#4
9937.84
Supplier#000005969
ROMANIA
108438
Manufacturer#1
9936.22
Supplier#000005250
UNITED KINGDOM
249
Manufacturer#4
9923.77
Supplier#000002324
GERMANY
29821
Manufacturer#4
9871.22
Supplier#000006373
GERMANY
43868
Manufacturer#5
[90 more rows]
7887.08
Supplier#000009792
GERMANY
164759
Manufacturer#3
7871.50
Supplier#000007206
RUSSIA
104695
Manufacturer#1
7852.45
Supplier#000005864
RUSSIA
8363
Manufacturer#4
Page 31
7850.66
Supplier#000001518
UNITED KINGDOM
86501
Manufacturer#1
7843.52
Supplier#000006683
FRANCE
11680
Manufacturer#4
S_ADDRESS
S_PHONE
S_COMMENT
QKuHYh,vZGiwu2FWEJo
LDx04
33-429-790-6131
uriously regular requests hag
ANDENSOSmk,miq23Xfb5R
Wt6dvUcvt6Qa
29-520-692-3537
efully express instructions. regular requests

against the slyly fin
B3rqp0xbSEim4Mpy
2RH J
33-320-228-2957
etect about the furiously final accounts. slyly

ironic pinto beans sleep inside
the furiously
y3OD9UywSTOk
17-779-299-1839
ackages boost blithely. blithely regular deposits c
J8fcXWsTqM
17-813-485-8637
etect blithely bold asymptotes. fluffily ironic

platelets wake furiously; blit
Y28ITVeYriT3kIGd
V2K8fSZ
V2UqT5H1Otz
17-988-938-4296
ckly around the carefully fluffy theodolites.

slyly ironic pack
3w
fNCnrVmvJjE95sgW
ZzvW
32-432-452-7731
ironic requests. furiously final theodolites

cajole. final, express packages sle
ep. quickly reg
WCNfBPZeSXh3h,c
32-454-883-3821
usly unusual pinto beans. brave ideas sleep carefully quickly ironi
ONda3YJiHKJOC
33-730-383-3892
ifts haggle fluffily pending pai
2Z0JGkiv01Y00oCF
wUGfviIbhzCdy
16-464-517-8943
express, final pinto beans x-ray slyly asymptotes. unusual, unusual
[90 More Rows]
Page 32
3.6
Shipping Priority Query (Q3)

This query retrieves the 10 unshipped orders with the highest value.
3.6.1
Business Question
The Shipping Priority Query retrieves the shipping priority and potential revenue, defined as the sum of
l_extendedprice * (1-l_discount), of the orders having the largest revenue among those that had not been shipped as
of a given date. Orders are listed in decreasing order of revenue. If more than 10 unshipped orders exist, only the 10
orders with the largest revenue are listed.
3.6.2

select
l_orderkey,
sum(l_extendedprice*(1-l_discount)) as revenue,
o_orderdate,
o_shippriority
from
customer,
orders,
lineitem
where
c_mktsegment = '[SEGMENT]'
and c_custkey = o_custkey
and l_orderkey = o_orderkey
and o_orderdate < date '[DATE]'
and l_shipdate > date '[DATE]'
group by
l_orderkey,
o_orderdate,
o_shippriority
order by
revenue desc,
o_orderdate;
3.6.3
Values for the following substitution parameters must be generated and used to build the executable query text:
1. SEGMENT is randomly selected within the list of values defined for Segments in Clause 5.2.2.13;
2. DATE is a randomly selected day within [1995-03-01 .. 1995-03-31].
3.6.4
Query Validation
Page 33
1. SEGMENT = BUILDING;
2. DATE = 1995-03-15.
L_ORDERKEY
REVENUE
O_ORDERDATE
O_SHIPPRIORITY
2456423
406181.01
1995-03-05
3459808
405838.70
1995-03-04
492164
390324.06
1995-02-19
1188320
384537.94
1995-03-09
2435712
378673.06
1995-02-26
4878020
378376.80
1995-03-12
5521732
375153.92
1995-03-13
2628192
373133.31
1995-02-22
993600
371407.46
1995-03-05
2300070
367371.15
1995-03-13
Page 34
3.7
Order Priority Checking Query (Q4)

This query determines how well the order priority system is working and gives an assessment of customer satisfaction.
3.7.1
Business Question
The Order Priority Checking Query counts the number of orders ordered in a given quarter of a given year in which
at least one lineitem was received by the customer later than its committed date. The query lists the count of such
orders for each order priority sorted in ascending priority order.
3.7.2
select
o_orderpriority,
count(*) as order_count
from orders
where
o_orderdate >= date '[DATE]'
and o_orderdate < date '[DATE]' + interval '3' month
and exists (
select
*
from
lineitem
where
l_orderkey = o_orderkey
and l_commitdate < l_receiptdate
)
group by
o_orderpriority
order by
o_orderpriority;
3.7.3
1. DATE is the first day of a randomly selected month between the first month of 1993 and the 10th month of
1997.
3.7.4
Query Validation
1. DATE = 1993-07-01.
Page 35
O_ORDERPRIORITY
ORDER_COUNT
1-URGENT
10594
2-HIGH
10476
3-MEDIUM
10410
4-NOT SPECIFIED
10556
5-LOW
10487
Page 36
3.8
Local Supplier Volume Query (Q5)

This query lists the revenue volume done through local suppliers.
3.8.1
Business Question
The Local Supplier Volume Query lists for each nation in a region the revenue volume that resulted from lineitem
transactions in which the customer ordering parts and the supplier filling them were both within that nation. The
query is run in order to determine whether to institute local distribution centers in a given region. The query considers only parts ordered in a given year. The query displays the nations and revenue volume in descending order by
revenue. Revenue volume for all qualifying lineitems in a particular nation is defined as sum(l_extendedprice * (1 l_discount)).
3.8.2
select
n_name,
sum(l_extendedprice * (1 - l_discount)) as revenue
from
customer,
orders,
lineitem,
supplier,
nation,
region
where
c_custkey = o_custkey
and l_suppkey = s_suppkey
and c_nationkey = s_nationkey
and o_orderdate >= date '[DATE]'
and o_orderdate < date '[DATE]' + interval '1' year
group by
n_name
order by
revenue desc;
3.8.3
1. REGION is randomly selected within the list of values defined for R_NAME in Clause 5.2.3;
2. DATE is the first of January of a randomly selected year within [1993 .. 1997].
3.8.4
Query Validation
Page 37

1. REGION = ASIA;
2. DATE = 1994-01-01.
N_NAME
REVENUE
INDONESIA
55502041.17
VIETNAM
55295087.00
CHINA
53724494.26
INDIA
52035512.00
JAPAN
45410175.70
Page 38
3.9
Forecasting Revenue Change Query (Q6)

This query quantifies the amount of revenue increase that would have resulted from eliminating certain companywide discounts in a given percentage range in a given year. Asking this type of "what if" query can be used to look
for ways to increase revenues.
3.9.1
Business Question
The Forecasting Revenue Change Query considers all the lineitems shipped in a given year with discounts between
DISCOUNT-0.01 and DISCOUNT+0.01. The query lists the amount by which the total revenue would have
increased if these discounts had been eliminated for lineitems with l_quantity less than quantity. Note that the
potential revenue increase is equal to the sum of [l_extendedprice * l_discount] for all lineitems with discounts and
quantities in the qualifying range.
3.9.2
select
sum(l_extendedprice*l_discount) as revenue
from
lineitem
where
l_shipdate >= date '[DATE]'
and l_shipdate < date '[DATE]' + interval '1' year
and l_discount between [DISCOUNT] - 0.01 and [DISCOUNT] + 0.01
and l_quantity < [QUANTITY];
3.9.3
1. DATE is the first of January of a randomly selected year within [1993 .. 1997];
2. DISCOUNT is randomly selected within [0.02 .. 0.09];
3. QUANTITY is randomly selected within [24 .. 25].
3.9.4
Query Validation
1. DATE = 1994-01-01;
2. DISCOUNT = 0.06;
3. QUANTITY = 24.
REVENUE
Page 39
123141078.23
Page 40
3.10
Volume Shipping Query (Q7)

This query determines the value of goods shipped between certain nations to help in the re-negotiation of shipping
contracts.
3.10.1
Business Question
The Volume Shipping Query finds, for two given nations, the gross discounted revenues derived from lineitems in
which parts were shipped from a supplier in either nation to a customer in the other nation during 1995 and 1996.
The query lists the supplier nation, the customer nation, the year, and the revenue from shipments that took place in
that year. The query orders the answer by Supplier nation, Customer nation, and year (all ascending).
3.10.2
select
supp_nation,
cust_nation,
l_year, sum(volume) as revenue
from (
select
n1.n_name as supp_nation,
n2.n_name as cust_nation,
extract(year from l_shipdate) as l_year,
l_extendedprice * (1 - l_discount) as volume
from
supplier,
lineitem,
orders,
customer,
nation n1,
nation n2
where
s_suppkey = l_suppkey
and o_orderkey = l_orderkey
and s_nationkey = n1.n_nationkey
and c_nationkey = n2.n_nationkey
and (
(n1.n_name = '[NATION1]' and n2.n_name = '[NATION2]')
or (n1.n_name = '[NATION2]' and n2.n_name = '[NATION1]')
)
and l_shipdate between date '1995-01-01' and date '1996-12-31'
Page 41
) as shipping
group by
supp_nation,
cust_nation,
l_year
order by
supp_nation,
cust_nation,
l_year;
3.10.3
1. NATION1 is randomly selected within the list of values defined for N_NAME in Clause 5.2.3;
2. NATION2 is randomly selected within the list of values defined for N_NAME in Clause 5.2.3 and must be different from the value selected for NATION1 in item 1 above.
3.10.4
Query Validation
1. NATION1 = FRANCE;
2. NATION2 = GERMANY.
SUPP_NATION
CUST_NATION
YEAR
REVENUE
FRANCE
GERMANY
1995
54639732.73
FRANCE
GERMANY
1996
54633083.31
GERMANY
FRANCE
1995
52531746.67
GERMANY
FRANCE
1996
52520549.02
Page 42
3.11
National Market Share Query (Q8)

This query determines how the market share of a given nation within a given region has changed over two years for
a given part type.
3.11.1
Business Question
The market share for a given nation within a given region is defined as the fraction of the revenue, the sum of
[l_extendedprice * (1-l_discount)], from the products of a specified type in that region that was supplied by suppliers from the given nation. The query determines this for the years 1995 and 1996 presented in this order.
3.11.2
select
o_year,
sum(case
when nation = '[NATION]'
then volume
else 0
end) / sum(volume) as mkt_share
from (
select
extract(year from o_orderdate) as o_year,
l_extendedprice * (1-l_discount) as volume,
n2.n_name as nation
from
part,
supplier,
lineitem,
orders,
customer,
nation n1,
nation n2,
region
where
p_partkey = l_partkey
and s_suppkey = l_suppkey
and o_custkey = c_custkey
and n1.n_regionkey = r_regionkey
and o_orderdate between date '1995-01-01' and date '1996-12-31'
and p_type = '[TYPE]'
Page 43
) as all_nations
group by
o_year
order by
o_year;
3.11.3
1. NATION is randomly selected within the list of values defined for N_NAME in Clause 5.2.3;
2. REGION is the value defined in Clause 5.2.3 for R_NAME where R_REGIONKEY corresponds to
N_REGIONKEY for the selected NATION in item 1 above;
3. TYPE is randomly selected within the list of 3-syllable strings defined for Types in Clause 5.2.2.13.
3.11.4
Query Validation
1. NATION = BRAZIL;
2. REGION = AMERICA;
3. TYPE = ECONOMY ANODIZED STEEL.
YEAR
MKT_SHARE
1995
.03
1996
.04
Page 44
3.12
Product Type Profit Measure Query (Q9)

This query determines how much profit is made on a given line of parts, broken out by supplier nation and year.
3.12.1
Business Question
The Product Type Profit Measure Query finds, for each nation and each year, the profit for all parts ordered in that
year that contain a specified substring in their names and that were filled by a supplier in that nation. The profit is
defined as the sum of [(l_extendedprice*(1-l_discount)) - (ps_supplycost * l_quantity)] for all lineitems describing
parts in the specified line. The query lists the nations in ascending alphabetical order and, for each nation, the year
and profit in descending order by year (most recent first).
3.12.2
select
nation,
o_year,
sum(amount) as sum_profit
from (
select
n_name as nation,
l_extendedprice * (1 - l_discount) - ps_supplycost * l_quantity as amount
from
part,
supplier,
lineitem,
partsupp,
orders,
nation
where
s_suppkey = l_suppkey
and ps_suppkey = l_suppkey
and ps_partkey = l_partkey
and p_partkey = l_partkey
and p_name like '%[COLOR]%'
) as profit
group by
nation,
o_year
order by
nation,
o_year desc;
3.12.3
1. COLOR is randomly selected within the list of values defined for the generation of P_NAME in Clause 5.2.3.
Page 45
3.12.4
Query Validation
1. COLOR = green.
NATION
YEAR
SUM_PROFIT
ALGERIA
1998
31342867.24
ALGERIA
1997
57138193.03
ALGERIA
1996
56140140.13
ALGERIA
1995
53051469.66
ALGERIA
1994
53867582.12
VIETNAM
1996
50488161.42
VIETNAM
1995
49658284.61
VIETNAM
1994
50596057.26
VIETNAM
1993
50953919.14
VIETNAM
1992
49613838.33
[165 more rows]
Page 46
3.13
Returned Item Reporting Query (Q10)

The query identifies customers who might be having problems with the parts that are shipped to them.
3.13.1
Business question
The Returned Item Reporting Query finds the top 20 customers, in terms of their effect on lost revenue for a given
quarter, who have returned parts. The query considers only parts that were ordered in the specified quarter. The
query lists the customer's name, address, nation, phone number, account balance, comment information and revenue
lost. The customers are listed in descending order of lost revenue. Revenue lost is defined as
sum(l_extendedprice*(1-l_discount)) for all qualifying lineitems.
3.13.2

select
c_custkey,
c_name,
sum(l_extendedprice * (1 - l_discount)) as revenue,
c_acctbal,
n_name,
c_address,
c_phone,
c_comment
from
customer,
orders,
lineitem,
nation
where
and o_orderdate >= date '[DATE]'
and o_orderdate < date '[DATE]' + interval '3' month
and l_returnflag = 'R'
and c_nationkey = n_nationkey
group by
c_custkey,
c_name,
c_acctbal,
c_phone,
n_name,
c_address,
c_comment
order by
revenue desc;
3.13.3
Page 47
1994.
3.13.4
Query Validation
1. DATE = 1993-10-01.
C_CUSTKEY
C_NAME
REVENUE
C_ACCTBAL
N_NAME
57040
Customer#000057040
734235.24
632.87
JAPAN
143347
Customer#000143347
721002.70
2557.47
EGYPT
60838
Customer#000060838
679127.31
2454.77
BRAZIL
101998
Customer#000101998
637029.57
3790.89
UNITED KINGDOM
125341
Customer#000125341
633508.09
4983.51
GERMANY
[10 more rows]
110246
Customer#000110246
566842.98
7763.35
VIETNAM
142549
Customer#000142549
563537.24
5085.99
INDONESIA
146149
Customer#000146149
557254.99
1791.55
ROMANIA
52528
Customer#000052528
556397.35
551.79
ARGENTINA
23431
Customer#000023431
554269.54
3381.86
ROMANIA
C_ADDRESS
C_PHONE
C_COMMENT
Eioyzjf4pp
22-895-641-3466
sits. slyly regular requests sleep alongside of

the regular inst
1aReFYv,Kw4
14-742-935-3718
ggle carefully enticing requests. final deposits

use bold, bold pinto beans. iro
nic, idle re
64EaJ5vMAHWJlBOxJ
klpNc2RJiWE
12-913-494-9813
01c9CILnNtfOQYmZj
33-593-865-6378
need to boost against the slyly regular account

ress foxes wake slyly after the bold excuses.
ironic platelets are furiously car
Page 48
S29ODD6bceU8QSuuE
JznkNaK
17-582-695-5962
arefully even depths. blithely even excuses

sleep furiously. foxes use except th
e dependencies. ca
7KzflgX MDOq7sOkI
31-943-426-9837
egular deposits serve blithely above the fl
ChqEoK43OysjdHbtK
Cp6dKqjNyvvi9
19-955-562-2398
sleep pending courts. ironic deposits against

the carefully unusual platelets ca
jole carefully express accounts.
s87fvzFQpU
29-744-164-6487
of the slyly silent accounts. quickly final

accounts across the
NFztyTOR10UOJ
11-208-192-3205
deposits hinder. blithely pending asymptotes

breach slyly regular re
HgiV0phqhaIa9aydN
oIlb
29-915-458-2654
nusual, even instructions: furiously stealthy n
[10 more rows]
Page 49
3.14
Important Stock Identification Query (Q11)

This query finds the most important subset of suppliers' stock in a given nation.
3.14.1
Business Question
The Important Stock Identification Query finds, from scanning the available stock of suppliers in a given nation, all
the parts that represent a significant percentage of the total value of all available parts. The query displays the part
number and the value of those parts in descending order of value.
3.14.2
select
ps_partkey,
sum(ps_supplycost * ps_availqty) as value
from
partsupp,
supplier,
nation
where
ps_suppkey = s_suppkey
and n_name = '[NATION]'
group by
ps_partkey having
sum(ps_supplycost * ps_availqty) > (
select
sum(ps_supplycost * ps_availqty) * [FRACTION]
from
partsupp,
supplier,
nation
where
ps_suppkey = s_suppkey
)
order by
value desc;
3.14.3
1. NATION is randomly selected within the list of values defined for N_NAME in Clause 5.2.3;
2. FRACTION is chosen as 0.0001 / SF.
3.14.4
Query Validation
Page 50

1. NATION = GERMANY;
2. FRACTION = 0.0001.
PS_PARTKEY
VALUE
129760
17538456.86
166726
16503353.92
191287
16474801.97
161758
16101755.54
34452
15983844.72
[1038 More Rows]
154731
7888301.33
101674
7879324.60
51968
7879102.21
72073
7877736.11
5182
7874521.73
Page 51
3.15
Shipping Modes and Order Priority Query (Q12)

This query determines whether selecting less expensive modes of shipping is negatively affecting the critical-priority orders by causing more parts to be received by customers after the committed date.
3.15.1
Business Question
The Shipping Modes and Order Priority Query counts, by ship mode, for lineitems actually received by customers
in a given year, the number of lineitems belonging to orders for which the l_receiptdate exceeds the l_commitdate
for two different specified ship modes. Only lineitems that were actually shipped before the l_commitdate are considered. The late lineitems are partitioned into two groups, those with priority URGENT or HIGH, and those with a
priority other than URGENT or HIGH.
3.15.2
select
l_shipmode,
sum(case
when o_orderpriority ='1-URGENT'
or o_orderpriority ='2-HIGH'
then 1
else 0
end) as high_line_count,
sum(case
when o_orderpriority <> '1-URGENT'
and o_orderpriority <> '2-HIGH'
then 1
else 0
end) as low_line_count
from
orders,
lineitem
where
o_orderkey = l_orderkey
and l_shipmode in ('[SHIPMODE1]', '[SHIPMODE2]')
and l_shipdate < l_commitdate
and l_receiptdate >= date '[DATE]'
and l_receiptdate < date '[DATE]' + interval '1' year
group by
l_shipmode
order by
l_shipmode;
3.15.3
1. SHIPMODE1 is randomly selected within the list of values defined for Modes in Clause 5.2.2.13;
2. SHIPMODE2 is randomly selected within the list of values defined for Modes in Clause 5.2.2.13 and must be
different from the value selected for SHIPMODE1 in item 1;
Page 52
3. DATE is the first of January of a randomly selected year within [1993 .. 1997].
3.15.4
Query Validation
1. SHIPMODE1 = MAIL;
2. SHIPMODE2 = SHIP;
3. DATE = 1994-01-01.
L_SHIPMODE
HIGH_LINE_COUNT
LOW_LINE_COUNT
MAIL
6202
9324
SHIP
6200
9262
Page 53
3.16
Customer Distribution Query (Q13)

This query seeks relationships between customers and the size of their orders.
3.16.1
Business Question
This query determines the distribution of customers by the number of orders they have made, including customers
who have no record of orders, past or present. It counts and reports how many customers have no orders, how many
have 1, 2, 3, etc. A check is made to ensure that the orders counted do not fall into one of several special categories
of orders. Special categories are identified in the order comment column by looking for a particular pattern.
3.16.2
select
c_count, count(*) as custdist
from (
select
c_custkey,
count(o_orderkey)
from
customer left outer join orders on
and o_comment not like %[WORD1]%[WORD2]%
group by
c_custkey
)as c_orders (c_custkey, c_count)
group by
c_count
order by
custdist desc,
c_count desc;
3.16.3
1. WORD1 is randomly selected from 4 possible values: special, pending, unusual, express.
2. WORD2 is randomly selected from 4 possible values: packages, requests, accounts, deposits.
3.16.4
Query Validation
For validation against the qualification database the query must be executed using the following substitution parameters and must produce the following output data:
1. WORD1 = special.
2. WORD2 = requests.
C_COUNT
CUSTDIST
Page 54
50005
6641
10
6532
11
6014
5937
[32 more rows]
38
37
40
41
39
Page 55
3.17
Promotion Effect Query (Q14)

This query monitors the market response to a promotion such as TV advertisements or a special campaign.
3.17.1
Business Question
The Promotion Effect Query determines what percentage of the revenue in a given year and month was derived
from promotional parts. The query considers only parts actually shipped in that month and gives the percentage.
Revenue is defined as (l_extendedprice * (1-l_discount)).
3.17.2
select
100.00 * sum(case
when p_type like 'PROMO%'
then l_extendedprice*(1-l_discount)
else 0
end) / sum(l_extendedprice * (1 - l_discount)) as promo_revenue
from
lineitem,
part
where
l_partkey = p_partkey
and l_shipdate >= date '[DATE]'
and l_shipdate < date '[DATE]' + interval '1' month;
3.17.3
1. DATE is the first day of a month randomly selected from a random year within [1993 .. 1997].
3.17.4
Query Validation
1. DATE = 1995-09-01.
PROMO_REVENUE
16.38
Page 56
3.18
Top Supplier Query (Q15)

This query determines the top supplier so it can be rewarded, given more business, or identified for special recognition.
3.18.1
Business Question
The Top Supplier Query finds the supplier who contributed the most to the overall revenue for parts shipped during
a given quarter of a given year. In case of a tie, the query lists all suppliers whose contribution was equal to the
maximum, presented in supplier number order.
3.18.2
create view revenue[STREAM_ID] (supplier_no, total_revenue) as

select
l_suppkey,
sum(l_extendedprice * (1 - l_discount))
from
lineitem
where
and l_shipdate < date '[DATE]' + interval '3' month
group by
l_suppkey;
select
s_suppkey,
s_name,
s_address,
s_phone,
total_revenue
from
supplier,
revenue[STREAM_ID]
where
s_suppkey = supplier_no
and total_revenue = (
select
max(total_revenue)
from
revenue[STREAM_ID]
)
order by
s_suppkey;
drop view revenue[STREAM_ID];
3.18.3
Page 57
1997.
3.18.4
Query Validation
1. DATE = 1996-01-01.
S_SUPPKEY
S_NAME
S_ADDRESS
S_PHONE
TOTAL_REVENUE
8449
Supplier#000008449
Wp34zim9qYFbVctdW
20-469-856-8873
1772627.21
Page 58
3.19
Parts/Supplier Relationship Query (Q16)

This query finds out how many suppliers can supply parts with given attributes. It might be used, for example, to
determine whether there is a sufficient number of suppliers for heavily ordered parts.
3.19.1
Business Question
The Parts/Supplier Relationship Query counts the number of suppliers who can supply parts that satisfy a particular
customer's requirements. The customer is interested in parts of eight different sizes as long as they are not of a given
type, not of a given brand, and not from a supplier who has had complaints registered at the Better Business Bureau.
Results must be presented in descending count and ascending brand, type, and size.
3.19.2
select
p_brand,
p_type,
p_size,
count(distinct ps_suppkey) as supplier_cnt
from
partsupp,
part
where
and p_brand <> '[BRAND]'
and p_type not like '[TYPE]%'
and p_size in ([SIZE1], [SIZE2], [SIZE3], [SIZE4], [SIZE5], [SIZE6], [SIZE7], [SIZE8])
and ps_suppkey not in (
select
s_suppkey
from
supplier
where
s_comment like '%Customer%Complaints%'
)
group by
p_brand,
p_type,
p_size
order by
supplier_cnt desc,
p_brand,
p_type,
p_size;
3.19.3
1. BRAND = Brand#MN where M and N are two single character strings representing two numbers randomly and
independently selected within [1 .. 5];
Page 59
2. TYPE is made of the first 2 syllables of a string randomly selected within the list of 3-syllable strings defined
for Types in Clause 5.2.2.13;
3. SIZE1 is randomly selected as a set of eight different values within [1 .. 50];
10. SIZE8 is randomly selected as a set of eight different values within [1 .. 50].
3.19.4
Query Validation
1. BRAND = Brand#45.
2. TYPE = MEDIUM POLISHED .
3. SIZE1 = 49
4. SIZE2 = 14
5. SIZE3 = 23
6. SIZE4 = 45
7. SIZE5 = 19
8. SIZE6 = 3
9. SIZE7 = 36
10. SIZE8 = 9.
P_BRAND
P_TYPE
P_SIZE
SUPPLIER_CNT
Brand#41
MEDIUM BRUSHED TIN
28
Brand#54
STANDARD BRUSHED COPPER
14
27
Brand#11
STANDARD BRUSHED TIN
23
24
Brand#11
STANDARD BURNISHED BRASS
36
24
Brand#15
MEDIUM ANODIZED NICKEL
24
Page 60
[18,304 more rows]
Brand#52
MEDIUM BRUSHED BRASS
45
Brand#53
MEDIUM BRUSHED TIN
45
Brand#54
ECONOMY POLISHED BRASS
Brand#55
PROMO PLATED BRASS
19
Brand#55
STANDARD PLATED TIN
49
Page 61
3.20
Small-Quantity-Order Revenue Query (Q17)

This query determines how much average yearly revenue would be lost if orders were no longer filled for small
quantities of certain parts. This may reduce overhead expenses by concentrating sales on larger shipments.
3.20.1
Business Question
The Small-Quantity-Order Revenue Query considers parts of a given brand and with a given container type and
determines the average lineitem quantity of such parts ordered for all orders (past and pending) in the 7-year database. What would be the average yearly gross (undiscounted) loss in revenue if orders for these parts with a quantity of less than 20% of this average were no longer taken?
3.20.2
select
sum(l_extendedprice) / 7.0 as avg_yearly
from
lineitem,
part
where
and p_brand = '[BRAND]'
and p_container = '[CONTAINER]'
and l_quantity < (
select
0.2 * avg(l_quantity)
from
lineitem
where
);
3.20.3
1. BRAND = 'Brand#MN' where MN is a two character string representing two numbers randomly and independently selected within [1 .. 5];
2. CONTAINER is randomly selected within the list of 2-syllable strings defined for Containers in Clause
5.2.2.13.
3.20.4
Query Validation
1. BRAND = Brand#23;
2. CONTAINER = MED BOX.
Page 62
AVG_YEARLY
348406.05
Page 63
3.21
Large Volume Customer Query (Q18)

The Large Volume Customer Query ranks customers based on their having placed a large quantity order. Large
quantity orders are defined as those orders whose total quantity is above a certain level.
3.21.1
Business Question
The Large Volume Customer Query finds a list of the top 100 customers who have ever placed large quantity
orders. The query lists the customer name, customer key, the order key, date and total price and the quantity for the
order.
3.21.2

select
c_name,
c_custkey,
o_orderkey,
o_orderdate,
o_totalprice,
sum(l_quantity)
from
customer,
orders,
lineitem
where
o_orderkey in (
select
l_orderkey
from
lineitem
group by
l_orderkey having
sum(l_quantity) > [QUANTITY]
)
group by
c_name,
c_custkey,
o_orderkey,
o_orderdate,
o_totalprice
order by
o_totalprice desc,
o_orderdate;
3.21.3
Page 64
1. QUANTITY is randomly selected within [312..315].

3.21.4
Query Validation
1. QUANTITY = 300
C_NAME
C_CUSTKEY
O_ORDERKEY
O_ORDERDATE
O_TOTALPRICE
Sum(L_QUANTITY)
Customer#000128120
128120
4722021
1994-04-07
544089.09
323.00
Customer#000144617
144617
3043270
1997-02-12
530604.44
317.00
Customer#000013940
13940
2232932
1997-04-13
522720.61
304.00
Customer#000066790
66790
2199712
1996-09-30
515531.82
327.00
Customer#000046435
46435
4745607
1997-07-03
508047.99
309.00
Customer#000069904
69904
1742403
1996-10-19
408513.00
305.00
Customer#000017746
17746
6882
1997-04-09
408446.93
303.00
Customer#000013072
13072
1481925
1998-03-15
399195.47
301.00
Customer#000082441
82441
857959
1994-02-07
382579.74
305.00
Customer#000088703
88703
2995076
1994-01-30
363812.12
302.00
[47 more rows]
Page 65
3.22
Discounted Revenue Query (Q19)

The Discounted Revenue Query reports the gross discounted revenue attributed to the sale of selected parts handled
in a particular manner. This query is an example of code such as might be produced programmatically by a data
mining tool.
3.22.1
Business Question
The Discounted Revenue query finds the gross discounted revenue for all orders for three different types of parts
that were shipped by air or delivered in person . Parts are selected based on the combination of specific brands, a list
of containers, and a range of sizes.
3.22.2
select
sum(l_extendedprice * (1 - l_discount) ) as revenue
from
lineitem,
part
where
(
and p_brand = [BRAND1]
and p_container in ( SM CASE, SM BOX, SM PACK, SM PKG)
and l_quantity >= [QUANTITY1] and l_quantity <= [QUANTITY1] + 10
and p_size between 1 and 5
and l_shipmode in (AIR, AIR REG)
and l_shipinstruct = DELIVER IN PERSON
)
or
(
and p_container in (MED BAG, MED BOX, MED PKG, MED PACK)
)
or
(
and p_container in ( LG CASE, LG BOX, LG PACK, LG PKG)
Page 66
);
3.22.3
1. QUANTITY1 is randomly selected within [1..10].
4. BRAND1, BRAND2, BRAND3 = 'Brand#MN' where each MN is a two character string representing two numbers randomly and independently selected within [1 .. 5]
3.22.4
Query Validation
1. QUANTITY1 = 1.
2. QUANTITY2 = 10.
3. QUANTITY3 = 20.
4. BRAND1 = Brand#12.
REVENUE
3083843.0578
Page 67
3.23
Potential Part Promotion Query (Q20)

The Potential Part Promotion Query identifies suppliers in a particular nation having selected parts that may be candidates for a promotional offer.
3.23.1
Business Question
The Potential Part Promotion query identifies suppliers who have an excess of a given part available; an excess is defined
to be more than 50% of the parts like the given part that the supplier shipped in a given year for a given nation. Only parts
whose names share a certain naming convention are considered.
3.23.2
select
s_name,
s_address
from
supplier, nation
where
s_suppkey in (
select
ps_suppkey
from
partsupp
where
ps_partkey in (
select
p_partkey
from
part
where
p_name like '[COLOR]%'
)
and ps_availqty > (
select
0.5 * sum(l_quantity)
from
lineitem
where
l_partkey = ps_partkey
and l_suppkey = ps_suppkey
and l_shipdate >= date('[DATE])
and l_shipdate < date('[DATE]) + interval 1 year
)
Page 68
)
order by
s_name;
3.23.3
1. COLOR is randomly selected within the list of values defined for the generation of P_NAME.
2. DATE is the first of January of a randomly selected year within 1993..1997.
3. NATION is randomly selected within the list of values defined for N_NAME in Clause 5.2.3.
3.23.4
Query Validation
1. COLOR = forest.
2. DATE = 1994-01-01.
3. NATION = CANADA.
S_NAME
S_ADDRESS
Supplier#000000020
iybAE,RmTymrZVYaFZva2SH,j
Supplier#000000091
YV45D7TkfdQanOOZ7q9QxkyGUapU1oOWU6q3
Supplier#000000197
YC2Acon6kjY3zj3Fbxs2k4Vdf7X0cd2F
Supplier#000000226
83qOdU2EYRdPQAQhEtn GRZEd
Supplier#000000285
Br7e1nnt1yxrw6ImgpJ7YdhFDjuBf
[194 more rows]

Supplier#000009862
rJzweWeN58
Supplier#000009868
ROjGgx5gvtkmnUUoeyy7v
Supplier#000009869
ucLqxzrpBTRMewGSM29t0rNTM30g1Tu3Xgg3mKag
Supplier#000009899
7XdpAHrzr1t,UQFZE
Supplier#000009974
7wJ,J5DKcxSU4Kp1cQLpbcAvB5AsvKT
Page 69
3.24
Suppliers Who Kept Orders Waiting Query (Q21)

This query identifies certain suppliers who were not able to ship required parts in a timely manner.
3.24.1
Business Question
The Suppliers Who Kept Orders Waiting query identifies suppliers, for a given nation, whose product was part of a multisupplier order (with current status of 'F') where they were the only supplier who failed to meet the committed delivery
date.
3.24.2
Return the first 100 selected rows.

select
s_name,
count(*) as numwait
from
supplier,
lineitem l1,
orders,
nation
where
s_suppkey = l1.l_suppkey
and o_orderkey = l1.l_orderkey
and o_orderstatus = 'F'
and l1.l_receiptdate > l1.l_commitdate
and exists (
select
*
from
lineitem l2
where
l2.l_orderkey = l1.l_orderkey
and l2.l_suppkey <> l1.l_suppkey
)
and not exists (
select
*
from
lineitem l3
where
l3.l_orderkey = l1.l_orderkey
and l3.l_suppkey <> l1.l_suppkey
and l3.l_receiptdate > l3.l_commitdate
Page 70
)
group by
s_name
order by
numwait desc,
s_name;
3.24.3
1. NATION is randomly selected within the list of values defined for N_NAME in Clause 5.2.3.
3.24.4
Query Validation
1. NATION = SAUDI ARABIA.
S_NAME
NUMWAIT
Supplier#000002829
20
Supplier#000005808
18
Supplier#000000262
17
Supplier#000000496
17
Supplier#000002160
17
[90 more rows]

Supplier#000001916
12
Supplier#000001925
12
Supplier#000002039
12
Supplier#000002357
12
Supplier#000002483
12
Page 71
3.25
Global Sales Opportunity Query (Q22)

The Global Sales Opportunity Query identifies geographies where there are customers who may be likely to make a
purchase.
3.25.1
Business Question
This query counts how many customers within a specific range of country codes have not placed orders for 7 years
but who have a greater than average positive account balance. It also reflects the magnitude of that balance.
Country code is defined as the first two characters of c_phone.
3.25.2
select
cntrycode,
count(*) as numcust,
sum(c_acctbal) as totacctbal
from (
select
substring(c_phone from 1 for 2) as cntrycode,
c_acctbal
from
customer
where
substring(c_phone from 1 for 2) in
('[I1]','[I2],'[I3]','[I4]','[I5]','[I6]','[I7]')
and c_acctbal > (
select
avg(c_acctbal)
from
customer
where
c_acctbal > 0.00
and substring (c_phone from 1 for 2) in
('[I1]','[I2]','[I3]','[I4]','[I5]','[I6]','[I7]')
)
and not exists (
select
*
from
orders
where
o_custkey = c_custkey
)
Page 72
) as custsale
group by
cntrycode
order by
cntrycode;
3.25.3
1. I1 I7 are randomly selected without repetition from the possible values for Country code as defined in Clause
5.2.2.9.
3.25.4
Query Validation
For validation against the qualification database the query must be executed using the following substitution parameters and must produce the following output data:
1. I1 = 13.
2. I2 = 31.
3. I3 = 23.
4. I4 = 29.
5. I5 = 30.
6. I6 = 18.
7. I7 = 17.
CNTRYCODE
NUMCUST
TOTACCTBAL
13
888
6737713.99
17
861
6460573.72
18
964
7236687.40
23
892
6701457.95
29
948
7158866.63
30
909
6808436.13
31
922
6806670.18
Page 73
3.26
General Requirements for Refresh functions
3.26.1
Refresh Function Overview

Each refresh function is defined by the following components:
3.26.2
The business rationale, which illustrates the business context in which the refresh functions could be used;
The refresh function definition, which defines in pseudo-code the function to be performed by the refresh
function;
The refresh data set, which defines the set of rows to be inserted or deleted by each execution of the refresh
function into or from the ORDERS and LINEITEM tables. This set of rows represents 0.1% of the initial population of these two tables (see Table 4: LINEITEM Cardinality).
Transaction Requirements for Refresh functions

The execution of each refresh function (RF1 or RF2) can be decomposed into any number of database transactions
as long as the following conditions are met:
All ACID properties are met;
Each atomic transaction includes a sufficient number of data modifications to maintain the logical database
consistency. For example, when adding or deleting a new order, the LINEITEM and the ORDERS tables are
both modified within the same transaction;
An output message is sent when the last transaction of the refresh function has completed successfully.
3.26.3
Refresh Function Compliance
3.26.3.1
The benchmark specification does not place any requirements on the implementation of the refresh functions other
than their functional equivalence to the refresh function definition and compliance with Clause 3.26.2. For RF1 and
RF2 only, the implementation is permitted to:
Use any language to write the code for the refresh functions;
Pre-process, compile and link the executable code on the SUT at any time prior to or during the measurement
interval;
Provide the SUT with the data to be inserted by RF1 or the set of keys for the rows to be deleted by RF2 prior
to the execution of the benchmark (this specifically does not allow pre-execution of the refresh functions).
Comment: The intent is to separate the resources required to generate the data to be inserted (or the set of key for
the rows to be deleted) from the resources required to execute insert and delete operations against the database.
Group the individual refresh functions into transactions and organize their execution serially or in parallel.
This grouping may be different in the power test and in the throughput test.
3.26.3.2
The refresh functions do not produce any output other than a message of successful completion.
3.26.3.3
The proper implementation of the refresh functions must be validated by the independent auditor who may request
additional tests to ascertain that the refresh functions execute in accordance with the benchmark requirements.
3.27
New Sales Refresh Function (RF1)

This refresh function adds new sales information to the database.
Page 74
3.27.1
Business Rationale
The New Sales refresh function inserts new rows into the ORDERS and LINEITEM tables in the database following the scaling and data generation methods used to populate the database.
3.27.2
Refresh Function Definition

LOOP (SF * 1500) TIMES
INSERT a new row into the ORDERS table
LOOP RANDOM(1, 7) TIMES
INSERT a new row into the LINEITEM table
END LOOP
END LOOP
Comment: The refresh functions can be implemented with much greater flexibility than the queries (see Clause
3.26.3). The definition provided here is an example only. Test sponsors may wish to explore other implementations.
3.27.3
Refresh Data Set

The set of rows to be inserted must be produced by DBGEN (or a modified version of DBGEN, see Clause 5.2.1)
using the -U option. This option will produce as many sets of rows as required for use in multi-stream tests.
3.28
Old Sales Refresh Function (RF2)

This refresh function removes old sales information from the database.
3.28.1
Business Rationale
The Old Sales refresh function removes rows from the ORDERS and LINEITEM tables in the database to emulate the
removal of stale or obsolete information.
3.28.2
Refresh Function Definition

LOOP (SF * 1500) TIMES
DELETE FROM ORDERS WHERE O_ORDERKEY = [value]
DELETE FROM LINEITEM WHERE L_ORDERKEY = [value]
END LOOP
Comment: The refresh functions can be implemented with much greater flexibility than the queries (see Clause
3.26.3). The definition provided here is an example only. Test sponsors may wish to explore other implementation
3.28.3
Refresh Data Set

The primary key values for the set of rows to be deleted must be produced by DBGEN (or a modified version of
DBGEN, see Clause 5.2.1) using the -U option. This option will produce as many sets of primary keys as required
for use in multi-stream throughput tests. The rows being deleted begin with the first row of each of the two targeted
tables.
3.29
Database Evolution Process

The test sponsor must assure the correctness of the database for each run within the performance test.
Page 75
This is accomplished by evolving the test database, keeping track of which set of inserted and deleted rows
should be used by RF1 and RF2 for each run (see Clause 6.1.1.4).
Comment: It is explicitly not permitted to rebuild or reload the test database during the performance test (see
Clause 6.1.1.3).
3.29.1
The test database may be endlessly reused if the test sponsor keeps careful track of how many pairs of refresh functions RF1/RF2 have been executed and completed successfully. For example, a test sponsor running five streams
would execute one RF1/RF2 pair during the power test using the first set of insert/delete rows produced by DBGEN
(see Clause 5.2.1). The throughput test would then execute the next five RF1/RF2 pairs using the second through
the sixth sets of inset/delete rows produced by DBGEN. The next run would use the sets of insert/delete rows produced by DBGEN for the seventh RF1/RF2 pair, and continue from there.
Page 76
4: DATABASE SYSTEM PROPERTIES

4.1
The ACID Properties
4.1.1
The ACID (Atomicity, Consistency, Isolation, and Durability) properties of transaction processing systems must be
supported by the system under test during the timed portion of this benchmark. Since TPC-H is not a transaction
processing benchmark, the ACID properties must be evaluated outside the timed portion of the test. It is the intent
of this section to informally define the ACID properties and to specify a series of tests that can be performed to
demonstrate that these properties are met.
4.1.2
While it is required for the system under test (SUT) to support the ACID properties defined in this Clause, the execution of the corresponding ACID tests is only required in lieu of supplying other sufficient evidence of the SUT's
support for these ACID properties. The existence of another published TPC-H benchmark for which support for the
ACID properties have been demonstrated using the tests defined in this Clause may be sufficient evidence that the
new SUT supports some or all of the required ACID properties. The determination of whether previously published
TPC-H test results are sufficient evidence of the above is left to the discretion of the auditor.
Comment 1: No finite series of tests can prove that the ACID properties are fully supported. Being able to pass the
specified tests is a necessary, but not sufficient, condition for meeting the ACID requirements.
Comment 2: The ACID tests are intended to demonstrate that the ACID properties are supported by the SUT and
enabled during the performance measurements. They are not intended to be an exhaustive quality assurance test.
4.1.3
The ACID tests must be performed against the qualification database. The same set of mechanisms used to ensure
full ACID properties of the qualification database during the ACID tests must be used/enabled for the test database
during the performance test. This applies both to attributes of the database and to attributes of the database session(s) used to execute the ACID and performance tests. The attributes of the session executing the ACID Query
(see Clause 4.1.6.3) must be the same as those used in the performance test query stream(s) (see Clause 6.1.2.3),
and the attributes of the session executing the ACID transaction (see Clause 4.1.6.2) must be the same as those used
in the performance test refresh stream (see Clause 6.1.2.4).
4.1.4
The mechanisms used to ensure durability of the qualification database must be enabled for the test database. For
example:
a)
If the qualification database relies on undo logs to ensure atomicity, then such logging must also be enabled
for the test database during the performance test, even though no transactions are aborted.
b)
If the qualification database relies on a database backup to meet the durability requirement (see Clause 4.5), a
backup must be taken of the test database.
c)
If the qualification database relies on RAID or mirroring to meet the durability requirement (see Clause 4.5),
these mechanisms must be active during the execution of the performance test.
4.1.5
The test sponsor must attest that the reported configuration would also pass the ACID tests with the test database.
4.1.6
The ACID Transaction and The ACID Query

Since this benchmark does not contain any OLTP transaction, a special ACID Transaction is defined for use in
some of the ACID tests. In addition, to simplify the demonstration that ACID properties are enabled while readonly queries are executing concurrently with other activities, a special ACID Query is defined.
4.1.6.1
Both the ACID transaction and the ACID Query utilize a truncation function to guarantee arithmetic function
portability and consistency of results. Define trunc(n,p) as
Page 77
trunc(
n, p) =
n * 10
10
which truncates n to the pth decimal place (e.g., trunc(1.357,2) = 1.35).

Comment: The intent of this clause is to specify the required functionality without dictating a particular implementation.
4.1.6.2
The ACID Transaction must be implemented to conform to the following transaction profile:
Given the set of input data (O_KEY, L_KEY, [delta]), with
O_KEY selected at random from the same distribution as that used to populate L_ORDERKEY in the qualification database (see Clause 5.2.3),
L_KEY selected at random from [1 .. M] where

M = SELECT MAX(L_LINENUMBER) FROM LINEITEM WHERE L_ORDERKEY = O_KEY
using the qualification database, and [delta] selected at random within [1 .. 100]:
BEGIN TRANSACTION
Read O_TOTALPRICE from ORDERS into [ototal] where O_ORDERKEY = [o_key]
Read L_QUANTITY, L_EXTENDEDPRICE, L_PARTKEY, L_SUPPKEY, L_TAX, L_DISCOUNT into
[quantity], [extprice], [pkey], [skey], [tax], [disc]
where L_ORDERKEY = [o_key] and L_LINENUMBER = [l_key]
Set [ototal] = [ototal] trunc( trunc([extprice] * (1 - [disc]), 2) * (1 + [tax]), 2)
Set [rprice] = trunc([extprice]/[quantity], 2)
Set [cost] = trunc([rprice] * [delta], 2)
Set [new_extprice] = [extprice] + [cost]
Set [new_ototal] = trunc([new_extprice] * (1.0 - [disc]), 2)
Set [new_ototal] = trunc([new_ototal] * (1.0 + [tax]), 2)
Set [new_ototal] = [ototal] + [new_ototal]
Update LINEITEM
where L_ORDERKEY = [o_key] and L_LINENUMBER = [l_key]
Set L_EXTENDEDPRICE = [new_extprice]
Set L_QUANTITY = [quantity] + [delta]
Write L_EXTENDEDPRICE, L_QUANTITY to LINEITEM
Update ORDERS where O_ORDERKEY = [o_key]
Set O_TOTALPRICE = [new_ototal]
Write O_TOTALPRICE to ORDERS
Insert Into HISTORY
Page 78
Values ([pkey], [skey], [o_key], [l_key], [delta], [current_date_time])

COMMIT TRANSACTION
Return [rprice], [quantity], [tax], [disc], [extprice], [ototal] to driver
Where HISTORY is a table required only for the ACID tests and defined as follows:
Column Name
H_P_KEY
identifier
Foreign reference to P_PARTKEY
H_S_KEY
identifier
Foreign reference to S_SUPPKEY
H_O_KEY
identifier
Foreign reference to O_ORDERKEY
H_L_KEY
integer
H_DELTA
integer
H_DATE_T
date and time to second
Comment: The values returned by the ACID Transaction are the old values, as read before the updates.
4.1.6.3
The ACID Query must be implemented to conform to the following functional query definition:
Given the input data:
O_KEY, selected within the same distributions as those used for the population of L_ORDERKEY in the
qualification database:
SELECT SUM(trunc(
trunc(L_EXTENDEDPRICE * (1 - L_DISCOUNT),2) * (1 + L_TAX),2))
FROM LINEITEM
WHERE L_ORDERKEY = [o_key]
4.1.6.4
The ACID Transaction and the ACID Query must be used to demonstrate that the ACID properties are fully supported by the system under test.
4.1.6.5
Although the ACID Transaction and the ACID Query do not involve all the tables of the TPC-H database, the
ACID properties must be supported for all tables of the TPC-H database.
4.2
Atomicity Requirements
4.2.1
Atomicity Property Definition

The system under test must guarantee that transactions are atomic; the system will either perform all individual
operations on the data, or will assure that no partially-completed operations leave any effects on the data.
4.2.2
Atomicity Tests
4.2.2.1
Perform the ACID Transaction (see Clause 4.1.5) for a randomly selected set of input data and verify that the appropriate rows have been changed in the ORDERS, LINEITEM, and HISTORY tables.
4.2.2.2
Perform the ACID Transaction for a randomly selected set of input data, substituting a ROLLBACK of the transaction for the COMMIT of the transaction. Verify that the appropriate rows have not been changed in the ORDERS,
LINEITEM, and HISTORY tables.
Page 79
4.3
Consistency Requirements
4.3.1
Consistency Property Definition

Consistency is the property of the application that requires any execution of transactions to take the database from
one consistent state to another.
4.3.2
Consistency Condition
4.3.2.1
A consistent state for the TPC-H database is defined to exist when:

O_TOTALPRICE =
SUM(trunc(trunc(L_EXTENDEDPRICE *(1 - L_DISCOUNT),2) * (1+L_TAX),2))
for each ORDERS and LINEITEM defined by (O_ORDERKEY = L_ORDERKEY)
4.3.2.2
A TPC-H database, when populated as defined in Clause 5.2, must meet the consistency condition defined in
Clause 4.3.2.1.
4.3.2.3
If data is replicated, as permitted under Clause 2.5.10, each copy must meet the consistency condition defined in
Clause 4.3.2.1.
4.3.3
Consistency Tests
To verify the consistency between the ORDERS, and LINEITEM tables, perform the following steps:
1. Verify that the ORDERS, and LINEITEM tables are initially consistent as defined in Clause 4.3.2.1, based on a
random sample of at least 10 distinct values of O_ORDERKEY.
2. Submit at least 100 ACID Transactions from each of at least the number of execution streams ( # query streams
+ 1 refresh stream) used in the reported throughput test (see Clause 6.3.4). Each transaction must use values of
(O_KEY, L_KEY, DELTA) randomly generated within the ranges defined in Clause 4.1.6.2. Ensure that all the
values of O_ORDERKEY chosen in Step 1 are used by some transaction in Step 2.
3. Re-verify the consistency of the ORDERS, and LINEITEM tables as defined in Clause 4.3.2.1 based on the
same sample values of O_ORDERKEY selected in Step 1.
4.4
Isolation Requirements
4.4.1
Isolation Property Definition

Isolation can be defined in terms of the following phenomena that may occur during the execution of concurrent
database transactions (i.e., read-write transactions or read-only queries):
P0 (Dirty Write): Database transaction T1 reads a data element and modifies it. Database transaction T2
then modifies or deletes that data element, and performs a COMMIT. If T1 were to attempt to reread the data element, it may receive the modified value from T2 or discover that the data element
has been deleted.
P1 (Dirty Read): Database transaction T1 modifies a data element. Database transaction T2 then reads that
data element before T1 performs a COMMIT. If T1 were to perform a ROLLBACK, T2 will have
read a value that was never committed and that may thus be considered to have never existed.
P2 (Non-repeatable Read): Database transaction T1 reads a data element. Database transaction T2 then
modifies or deletes that data element, and performs a COMMIT. If T1 were to attempt to re-read
Page 80
the data element, it may receive the modified value or discover that the data element has been
deleted.
P3 (Phantom): Database transaction T1 reads a set of values N that satisfy some <search condition>. Database transaction T2 then executes statements that generate one or more data elements that satisfy
the <search condition> used by database transaction T1. If database transaction T1 were to repeat
the initial read with the same <search condition>, it obtains a different set of values.
Each database transaction T1 and T2 above must be executed completely or not at all.
The following table defines four isolation levels with respect to the phenomena P0, P1, P2, and P3.
Phenomena P0
Phenomena P1
Phenomena P2
Phenomena P3
Level 0
Not Possible
Possible
Possible
Possible
Level 1
Not Possible
Not Possible
Possible
Possible
Level 2
Not Possible
Not Possible
Not Possible
Possible
Level 3
Not Possible
Not Possible
Not Possible
Not Possible
Table 1: Isolation Levels

The following terms are defined:
T1 = An instance of the ACID Transaction;
T2 = An instance of the ACID Transaction;
T3 = Any of the TPC-H queries 1 to 22 or an instance of the ACID query;
Tn = Any arbitrary transaction.
Although arbitrary, the transaction Tn shall not do dirty writes.
The following table defines the isolation requirements that must be met by TPC-H implementations.
Req. #
For transactions
this set:
1.
in
these phenomena:
must NOT be seen by this

transaction:
Textual Description:
{Ti, Tj} 1 i,j 2
P0, P1, P2, P3
Ti
Level 3 isolation between any two ACID

Transactions.
2.
{Ti, Tn}1 i 2
P0, P1, P2
Ti
Level 2 isolation for any ACID Transaction relative to any arbitrary transaction.
3.
{Ti, T3}1 i n
P0, P1
T3
Level 1 isolation for any of TPC-H queries 1 to 22 relative to any ACID Transaction and any arbitrary transaction.
Table 2: Isolation Requirements

Sufficient conditions must be enabled at either the system or application level to ensure the required isolation
defined above is obtained.
Page 81
However, the required isolation levels must not be obtained by the use of configurations or explicit session-level
options that give a particular session or transaction a priori exclusive access to the database.
The intent is not to preclude automatic mechanisms such as lock escalation, but to disallow configurations and
options that would a priori preclude queries and update transactions against the same database from making
progress concurrently.
In addition, the configuration of the database or session-level options must be such that the continuous submission
of arbitrary (read-only) queries against one or more tables could not indefinitely delay update transactions affecting
those tables from making progress.
4.4.2
Isolation Tests
For conventional locking schemes, isolation shall be tested as described below. Systems that implement other isolation schemes may require different validation techniques. It is the responsibility of the test sponsor to disclose those
techniques and the tests for them. If isolation schemes other than conventional locking are used, it is permissible to
implement these tests differently provided full details are disclosed.
The six tests described here are designed to verify that the system under test is configured to support the required
isolation levels, as defined in Clause 4.4.1. All Isolation Tests are performed using a randomly selected set of values
(P_KEY, S_KEY, O_KEY, L_KEY, DELTA).
Comment: In the isolation tests, the values returned by the ACID Transaction are the old values, as read before the
updates.
4.4.2.1
Isolation Test 1
This test demonstrates isolation for the read-write conflict of a read-write transaction and a read-only transaction
when the read-write transaction is committed. Perform the following steps:
1. Start an ACID Transaction Txn1 for a randomly selected O_KEY, L_KEY, and DELTA.
2. Suspend Txn1 immediately prior to COMMIT.
3. Start an ACID Query Txn2 for the same O_KEY as in Step 1. (Txn2 attempts to read the data that has just been
updated by Txn1.)
4. Verify that Txn2 does not see Txn1's updates.
5. Allow Txn1 to complete.
6. Txn2 should now have completed.
4.4.2.2
Isolation Test 2
This test demonstrates isolation for the read-write conflict of a read-write transaction and a read-only transaction
when the read-write transaction is rolled back. Perform the following steps:
1. Start an ACID Transaction Txn1 for a randomly selected O_KEY, L_KEY, and DELTA.
3. Start an ACID Query Txn2 for the same O_KEY as in Step 1. (Txn2 attempts to read the data that has just been
updated by Txn1.)
4. Verify that Txn2 does not see Txn1's updates.
Page 82
5. Force Txn1 to rollback.

6. Txn2 should now have completed.
4.4.2.3
Isolation Test 3
This test demonstrates isolation for the write-write conflict of two update transactions when the first transaction is
committed. Perform the following steps:
1. Start an ACID Transaction Txn1 for a randomly selected O_KEY, L_KEY, and DELTA1.
2. Stop Txn1 immediately prior to COMMIT.
3. Start another ACID Transaction Txn2 for the same O_KEY, L_KEY and for a randomly selected DELTA2.
(Txn2 attempts to read and update the data that has just been updated by Txn1.)
4. Verify that Txn2 waits.
5. Allow Txn1 to complete. Txn2 should now complete.
6. Verify that
Txn2.L_EXTENDEDPRICE = Txn1.L_EXTENDEDPRICE+
(DELTA1 * (Txn1.L_EXTENDEDPRICE / Txn1.L_QUANTITY))
4.4.2.4
Isolation Test 4
This test demonstrates isolation for the write-write conflict of two update transactions when the first transaction is
rolled back. Perform the following steps:
1. Start an ACID Transaction Txn1 for a randomly selected O_KEY, L_KEY, and DELTA1.
2. Stop Txn1 immediately prior to COMMIT.
3. Start another ACID Transaction Txn2 for the same O_KEY, L_KEY and for a randomly selected DELTA2.
(Txn2 attempts to read and update the data that has just been updated by Txn1.)
4. Verify that Txn2 waits.
5. Force Txn1 to rollback. Txn2 should now complete.
6. Verify that
Txn2.L_EXTENDEDPRICE = Txn1.L_EXTENDEDPRICE
4.4.2.5
Isolation Test 5
This test demonstrates the ability of read and write transactions affecting different database tables to make progress
concurrently.
1. Start an ACID Transaction Txn1 with randomly selected values of O_KEY, L_KEY and DELTA.
3. Start a transaction Txn2 that does the following:
4. Select random values of PS_PARTKEY and PS_SUPPKEY. Return all columns of the PARTSUPP table for
which PS_PARTKEY and PS_SUPPKEY are equal to the selected values.
5. Verify that Txn2 completes.
6. Allow Txn1 to complete. Verify that the appropriate rows in the ORDERS, LINEITEM and HISTORY tables
have been changed.
Page 83
4.4.2.6
Isolation Test 6
This test demonstrates that the continuous submission of arbitrary (read-only) queries against one or more tables of
the database does not indefinitely delay update transactions affecting those tables from making progress.
1. Start a transaction Txn1. Txn1 executes Q1 (from Clause 3.4) against the qualification database where the substitution parameter [delta] is chosen from the interval [0 .. 2159] so that the query runs for a sufficient length of
time.
Comment: Choosing [delta] = 0 will maximize the run time of Txn1.
2. Before Txn1 completes, submit an ACID Transaction Txn2 with randomly selected values of O_KEY, L_KEY
and DELTA.
If Txn2 completes before Txn1 completes, verify that the appropriate rows in the ORDERS, LINEITEM and HISTORY tables have been changed. In this case, the test is complete with only Steps 1 and 2. If Txn2 will not complete before Txn1 completes, perform Steps 3 and 4:
3. Ensure that Txn1 is still active. Submit a third transaction Txn3, which executes Q1 against the qualification
database with a test-sponsor selected value of the substitution parameter [delta] that is not equal to the one used
in Step 1.
4. Verify that Txn2 completes before Txn3, and that the appropriate rows in the ORDERS, LINEITEM and HISTORY tables have been changed.
Comment: In some implementations Txn2 will not queue behind Txn1. If Txn2 completes prior to Txn1 completion, it is not necessary to run Txn3 in order to demonstrate that updates will be processed in a timely manner as
required by Isolation Tests.
4.5
Durability Requirements
The SUT must guarantee durability: the ability to preserve the effects of committed transactions and ensure database consistency after recovery from any one of the failures listed in Clause 4.5.3.
Comment: No system provides complete durability (i.e., durability under all possible types of failures). The specific set of single failures addressed in Clause 4.5.3 is deemed sufficiently significant to justify demonstration of
durability across such failures.
4.5.1
Durable Medium Definition

A durable medium is a data storage medium that is either:
a)
An inherently non-volatile medium (e.g., magnetic disk, magnetic tape, optical disk, etc.) or;
b)
A volatile medium with its own self-contained power supply that will retain and permit the transfer of data,
before any data is lost, to an inherently non-volatile medium after the failure of external power.
A configured and priced Uninterruptible Power Supply (UPS) is not considered external power.
Comment: A durable medium can fail; this is usually protected against by replication on a second durable medium
(e.g., mirroring) or logging to another durable medium. Memory can be considered a durable medium if it can preserve data long enough to satisfy the requirement (b) above, for example, if it is accompanied by an Uninterruptible
Power Supply, and the contents of memory can be transferred to an inherently non-volatile medium during the failure. Note that no distinction is made between main memory and memory performing similar permanent or temporary data storage in other parts of the system (e.g., disk controller caches).
Page 84
4.5.2
Committed Property Definition
4.5.2.1
A transaction is considered committed when the transaction manager component of the system has either written
the log or written the data for the committed updates associated with the transaction to a durable medium.
Comment 1: Transactions can be committed without the user subsequently receiving notification of that fact, since
message integrity is not required for TPC-H.
Comment 2: Although the order of operations in the ACID Transaction is immaterial, the actual return of data cannot begin until the commit operation has successfully completed.
4.5.2.2
To facilitate the execution of the durability tests the driver must maintain a durable success file that records the
details of each transaction which has successfully completed and whose message has been returned to the driver. At
the time of an induced failure this success file must contain a record of all transactions which have been committed,
except for transactions whose commit notification message to the driver was interrupted by the failure.
The durability success file is required only for the durability tests and must contain the following fields:
Fields
Datatype Definition
P_KEY
identifierForeign reference to P_PARTKEY
S_KEY
identifierForeign reference to S_SUPPKEY
O_KEY
identifierForeign reference to O_ORDERKEY
L_KEY
integer
DELTA
integer
DATE_T
date and time to second
Comment: If the driver resides on the SUT, the success file must be isolated from the TPC-H database. For example, the success file must be written outside of the ACID Transaction, and if the durability of the success file is provided by the same data manager as the TPC-H database, it must use a different log file.
4.5.3
Durability Across Single Failures

The test sponsor is required to guarantee that the test system will preserve the database and the effects of committed
updates after recovery from any of the failures listed below:
Permanent irrecoverable failure of any single durable medium containing TPC-H database tables or recovery
log data. The media to be failed is to be chosen at random by the auditor, and cannot be specially prepared.
Comment: If main memory is used as a durable medium, then it must be considered as a potential single point of
failure. Sample mechanisms to survive single durable medium failures are database archiving in conjunction with a
redo (after image) log, and mirrored durable media. If memory is the durable medium and mirroring is the mechanism used to ensure durability, then the mirrored memories must be independently powered.
Instantaneous interruption (system crash/system hang) in processing which requires system re-boot to recover.
Comment: This implies abnormal system shutdown, which requires loading of a fresh copy of the operating system
from the boot device. It does not necessarily imply loss of volatile memory. When the recovery mechanism relies on
the pre-failure contents of volatile memory, the means used to avoid the loss of volatile memory (e.g., an Uninterruptible Power Supply) must be included in the system cost calculation. A sample mechanism to survive an instantaneous interruption in processing is an undo/redo log.
Failure of all or part of memory (loss of contents).
Page 85
Comment: This implies that all or part of memory has failed. This may be caused by a loss of external power or the
permanent failure of a memory board.
SUT Power Failure: Loss of all external power to the SUT for an indefinite time period.
Comment: To demonstrate durability in a cluster during a power failure, the largest subset of the SUT maintained
by a single UPS must be failed. For example, if a system has one UPS per node or set of nodes, it is sufficient to fail
one node or that set of nodes. If there is only one UPS for the entire system, then the entire system must be failed. In
either case, all UPSs must be priced.
Regardless of UPS configuration, at least one node of each subset of the nodes in the cluster providing a distinct
function must be failed.
4.5.4
Durability Tests
The intent of these tests is to demonstrate that all transactions whose output messages have been received by the
driver have in fact been committed in spite of any single failure from the list in Clause 4.5.3 and that all consistency
conditions are still met after the database is recovered.
For each of the failure types defined in Clause 4.5.3 perform the following steps:
1. Verify that the ORDERS, and LINEITEM tables are initially consistent as defined in Clause 4.3.2.1, based on a
random sample of at least 10 distinct values of O_ORDERKEY.
2. Asynchronously submit a minimum of 200 ACID Transactions from each of at least the number of the execution streams (# query streams + 1 refresh stream) used in the reported throughput test. Each transaction must use
values of (O_KEY, L_KEY, DELTA) randomly generated within the ranges defined in Clause 4.1.6.2. Ensure
that all the values of O_ORDERKEY chosen in Step 1 are used by some transaction in Step 2. It must be demonstrated that transactions are in progress at the time of the failure.
3. Wait until at least 100 of the ACID transactions from each stream submitted in Step 2 have completed. Cause
the failure selected from the list in Clause 4.5.3. At the time of the failure, it must be demonstrated that:
At least one transaction is in flight.
No stream has completed submission of the required minimum number of ACID transactions as defined in
Step 2.
Comment: The intent is that the failure is induced while all streams are continuously submitting and executing
transactions. If the number of in-flight transactions at the point of failure is less than the number of streams, this is
assumed to be a random consequence of interrupting some streams during the very small interval between committing one transaction and submitting the next.
4. Restart the system under test using normal recovery procedures.
5. Compare the contents of the durability success file and the HISTORY table to verify that records in the success
file for a committed ACID Transaction have a corresponding record in the HISTORY table and that no success
record exists for uncommitted transactions. Count the number of entries in the success file and in the HISTORY
table and report any difference.
Comment: This difference can only be due to transactions that were committed on the system under test, but for
which the data was not written in the success file before the failure.
6. Re-verify the consistency of the ORDERS, and LINEITEM tables as defined in Clause 4.3.2.1.
Page 86
5: SCALING AND DATABASE POPULATION

5.1
Database Definition and Scaling
5.1.1
Test Database
5.1.1.1
The test database is the database used to execute the load test and the performance test (see Clause 6.1.1.4).
5.1.1.2
The test database must be scaled as defined in Clause 5.1.3.
5.1.1.3
The test database must be populated according to Clause 5.2.
5.1.2
Qualification Database
5.1.2.1
A qualification database must be created and populated for use in the query validation test described in Clause 3.3.
The intent is that the functionality exercised by running the validation queries against the qualification database be
the same as that exercised against the test database during the performance test. To this end, the qualification database must be identical to the test database in virtually every regard except size, including but not limited to:
Column definitions;
Method of data generation and loading;
Statistics gathering method;
ACID property implementation;
Type of partitioning (but not degree of partitioning);
Replication
Table type (if there is a choice);
Auxiliary data structures (e.g., indices).
The qualification database may differ from the test database only if the difference is directly related to the difference in sizes. For example, if the test database employs horizontal partitioning (see Clause 2.5.3), then the qualification database must also employ horizontal partitioning, though the number of partitions may differ in each case.
As another example, the qualification database could be configured such that it uses a representative sub-set of the
processors/cores/threads, memory and disks used by the test database configuration. If the qualification database
configuration differs from the test database configuration in any way, the differences must be disclosed (see Clause
9.3.5.8).
5.1.2.2
The population of the qualification database must be exactly equal to a scale factor, SF, of 1 (see Clause 5.1.3 for a
definition of SF).
5.1.3
Database Scaling Requirements
5.1.3.1
Scale factors used for the test database must be chosen from the set of fixed scale factors defined as follows:
1, 10, 30, 100, 300, 1000, 3000, 10000, 30000, 100000
The database size is defined with reference to scale factor 1 (i.e., SF = 1; approximately 1GB as per Clause 5.2.5),
the minimum required size for a test database. Therefore, the following series of database sizes corresponds to the
series of scale factors and must be used in the metric names QphH@Size and Price-per-QphH@Size (see Clause
6.4), as well as in the executive summary statement (see Appendix E):
Page 87
1GB, 10GB, 30GB, 100GB, 300GB, 1000GB, 3000GB, 10000GB, 30000GB, 100000GB
Where GB stands for gigabyte, defined to be 230 bytes.
Comment 1: Although the minimum size of the test database for a valid performance test is 1GB (i.e., SF = 1), a
test database of 3GB (i.e., SF = 3) is not permitted. This requirement is intended to encourage comparability of
results at the low end and to ensure a substantial actual difference in test database sizes.
Comment 2: The maximum size of the test database for a valid performance test is currently set at 100000 (i.e., SF
= 100,000). The TPC recognizes that additional benchmark development work is necessary to allow TPC-H to scale
beyond that limit.
5.1.3.2
Test sponsors must choose the database size they want to execute against by selecting a size and corresponding
scale factor from the defined series.
5.1.3.3
The ratio of total data storage to database size must be computed by dividing the total durable data storage of the
priced configuration (expressed in GB) by the size chosen for the test database as defined in Scale factors used for
the test database must be chosen from the set of fixed scale factors defined as follows:. The ratio must be reported to
the nearest 1/100th, rounded up. The ratio must be included in both the Full Disclosure Report and the Executive
Summary.
5.2
DBGEN and Database Population
5.2.1
The DBGEN Program
5.2.1.1
The test database and the qualification database must be populated with data that meets the requirements of Clause
5.2.2 and Clause 5.2.3. The DBGEN source code provided in Appendix D is a sample implementation of a database
population generator. It has been written in ANSI 'C' and has been ported to a large number of platforms. If DBGEN
is used, its version and release numbers must match the version and release numbers of the benchmark specification.
Comment: Use of DBGEN is strongly recommended. Exact query answer set compliance is required and this may
not be possible unless data is identical to DBGEN's output. This, in turn, may not be possible without duplicating
the exact sequence of random numbers generated by DBGEN.
5.2.1.2
The data generated by DBGEN are meant to be compliant with Clause 5.2.2 and Clause 5.2.3. In case of differences
between the content of these two Clauses and the data generated by DBGEN, the latter prevails.
5.2.1.3
If a program other than DBGEN is used to populate the database, the resultant data must meet the following
requirements in order to be considered correct:
The content of individual columns must match that produced by DBGEN;
The data format of individual columns must be identical to that produced by DBGEN;
The number of rows generated for a given scale factor must match that produced by DBGEN.
Comment 1: The intent of this clause is to allow for modification of the DBGEN code required for portability or
speed, while precluding any change that effects the resulting data.
Comment 2: If a program other than DBGEN is used to populate the database, its complete source code must be
included in the full disclosure report. Alternatively, if a modified version of DBGEN is used, the modified source
files must be disclosed in full.
Page 88
5.2.2
Definition Of Terms
5.2.2.1
The term random means independently selected and uniformly distributed over the specified range of values.
5.2.2.2
The term unique within [x] represents any one value within a set of x values between 1 and x, unique within the
scope of rows being populated.
5.2.2.3
The notation random value [x .. y] represents a random value between x and y inclusively, with a mean of (x+y)/2,
and with the same number of digits of precision as shown. For example, [0.01 .. 100.00] has 10,000 unique values,
whereas [1..100] has only 100 unique values.
5.2.2.4
The notation random string [list_name] represents a string selected at random within the list of strings list_name
as defined in Clause 5.2.2.13. Each string must be selected with equal probability.
5.2.2.5
The notation text appended with digit [text, x] represents a string generated by concatenating the sub-string text,
the character "# ", and the sub-string representation of the number x.
5.2.2.6
The notation random a-string [x] represents a string of length x comprised of randomly generated alphanumeric
characters within a character set of at least 64 symbols.
5.2.2.7
The notation random v-string [x] represents a string comprised of randomly generated alphanumeric characters
within a character set of at least 64 symbols. The length of the string is a random value [0.4 x .. 1.6 x] rounded up to
the next integer.
5.2.2.8
The term date represents a string of numeric characters separated by hyphens and comprised of a 4 digit year, 2
digit month and 2 digit day of the month.
5.2.2.9
The term phone number represents a string of numeric characters separated by hyphens and generated as follows:
Let i be an index into the list of strings Nations (i.e., ALGERIA is 0, ARGENTINA is 1, etc., see Clause 5.2.3),
Let country_code be the sub-string representation of the number (i + 10),
Let local_number1 be random [100 .. 999],
The phone number string is obtained by concatenating the following sub-strings:
country_code, "-", local_number1, "-", local_number2, "-", local_number3
5.2.2.10
The term text string[x] represents a substring of a 300 MB string populated according to the peudo text grammar
defined in Clause 5.2.2.14. The length of the substring is a random value [0.4 x .. 1.6 x] rounded up to the next integer. The substring offset is randomly chosen.
5.2.2.11
The term text string[x] represents a string generated by the pseudo text grammar defined in Clause 5.2.2.14. The
length of the string is a random value [0.4 x .. 1.6 x] rounded up to the next integer. The output of the pseudo text
grammar is truncated at the selected length.
5.2.2.12
All dates must be computed using the following values:
STARTDATE = 1992-01-01
5.2.2.13
CURRENTDATE = 1995-06-17
ENDDATE = 1998-12-31
The following list of strings must be used to populate the database:

Page 89
List name:Types
Each string is generated by the concatenation of a variable length syllable selected at random from each of the three following
lists and separated by a single space (for a total of 150 combinations).
Syllable 1
Syllable 2
Syllable 3
STANDARD
ANODIZED
TIN
SMALL
BURNISHED
NICKEL
MEDIUM
PLATED
BRASS
LARGE
POLISHED
STEEL
ECONOMY
BRUSHED
COPPER
PROMO
List name:Containers
Each string is generated by the concatenation of a variable length syllable selected at random from each of the two following
lists and separated by a single space (for a total of 40 combinations).
Syllable 1
Syllable 2
SM
CASE
LG
BOX
MED
BAG
JUMBO
JAR
WRAP
PKG
PACK
CAN
DRUM
Page 90
List name:Segments
AUTOMOBILE
BUILDING
FURNITURE
MACHINERY
2-HIGH
3-MEDIUM
4-NOT SPECIFIED
COLLECT COD
NONE
TAKE BACK RETURN
REG AIR
AIR
RAIL
SHIP
TRUCK
MAIL
FOB
foxes
ideas
theodolites
pinto beans
instructions
dependencies
excuses
platelets
asymptotes
courts
dolphins
multipliers
sauternes
warthogs
frets
dinos
attainments
somas
Tiresias'
patterns
forges
braids
hockey players
frays
warhorses
dugouts
notornis
epitaphs
pearls
tithes
waters
orbits
gifts
sheaves
depths
sentiments
decoys
realms
pains
grouches
HOUSEHOLD
List name:Priorities
1-URGENT
5-LOW
List name:Instructions
DELIVER IN PERSON
List name:Modes
List name:Nouns
escapades
List name:Verbs
Page 91
sleep
wake
are
cajole
haggle
nag
use
boost
affix
detect
integrate
maintain
nod
was
lose
sublate
solve
thrash
promise
engage
hinder
print
x-ray
breach
eat
grow
impress
mold
poach
serve
run
dazzle
snooze
doze
unwind
kindle
play
hang
believe
doubt
furious
sly
careful
blithe
quick
fluffy
slow
quiet
ruthless
thin
close
dogged
daring
brave
stealthy
permanent
enticing
idle
busy
regular
final
ironic
even
bold
sometimes
always
never
furiously
slyly
carefully
blithely
quickly
fluffily
slowly
quietly
ruthlessly
thinly
closely
doggedly
daringly
bravely
stealthily
permanently
enticingly
idly
busily
regularly
finally
ironically
evenly
boldly
silently
List name:Adjectives
silent
List name:Adverbs
Page 92
List name:Prepositions
about
above
according to
across
after
against
along
alongside of
among
around
at
atop
before
behind
beneath
beside
besides
between
beyond
by
despite
during
except
for
from
in place of
inside
instead of
into
near
of
on
outside
over
past
since
through
throughout
to
toward
under
until
up
upon
without
with
within
do
may
might
shall
will
would
can
could
should
ought to
must
will have to
shall have to
could have to
should have to
must have to
need to
try to
List name:Auxiliaries
List name:Terminators
--
5.2.2.14
Pseudo text used in the data population (see Clause 5.2.2.11) must conform to the following grammar:
text:<sentence>
|<text> <sentence>
;
sentence:<noun phrase> <verb phrase> <terminator>
|<noun phrase> <verb phrase> <prepositional phrase> <terminator>
|<noun phrase> <verb phrase> <noun phrase> <terminator>
|<noun phrase> <prepositional phrase> <verb phrase>
Page 93
<noun phrase> <terminator>

|<noun phrase> <prepositional phrase> <verb phrase>
<prepositional phrase> <terminator>
;
noun phrase:<noun>
|<adjective> <noun>
|<adjective>, <adjective> <noun>
|<adverb> <adjective> <noun>
;
verb phrase:<verb>
|<auxiliary> <verb>
|<verb> <adverb>
|<auxiliary> <verb> <adverb>
;
prepositional phrase: <preposition> the <noun phrase>
;
noun:selected from Nouns (as defined in Clause 5.2.2.13)
verb: selected from Verbs (as defined in Clause 5.2.2.13)
adjective: selected from Adjectives (as defined in Clause 5.2.2.13)
adverb: selected from Adverbs (as defined in Clause 5.2.2.13)
preposition: selected from Prepositions (as defined in Clause 5.2.2.13)
terminator: selected from Terminators (as defined in Clause 5.2.2.13)
auxiliary: selected from Auxiliary (as defined in Clause 5.2.2.13)
5.2.2.15
The grammar defined in Clause 5.2.2.14 relies on the weighted, non-uniform distribution of its constituent distributions (nouns, verbs, auxiliaries, etc.). A sponsor wishing to use a means of data generation other than DBGEN must
assure that the distributions defined in the DBGEN source code are reproduced exactly.
5.2.3
Test Database Data Generation

The data generated by DBGEN (see Clause 5.2.1) must be used to populate the database as follows (where SF is the
scale factor, see Clause 5.1.3.1):
SF * 10,000 rows in the SUPPLIER table with:

S_SUPPKEY unique within [SF * 10,000].
S_NAME text appended with digit ["Supplier", S_SUPPKEY].
S_ADDRESS random v-string[25].
S_NATIONKEY random value [0 .. 24].
S_PHONE generated according to Clause 5.2.2.9.
S_ACCTBAL random value [-999.99 .. 9,999.99].
S_COMMENT text string [63].
SF * 5 rows are randomly selected to hold at a random position a string matching "Customer%Complaints". Another SF * 5 rows are randomly selected to hold at a random position a string matching "Customer%Recommends", where % is a wildcard that denotes
zero or more characters.
SF * 200,000 rows in the PART table with:
Page 94
P_PARTKEY unique within [SF * 200,000].

P_NAME generated by concatenating five unique randomly selected strings from the following list, separated
by a single space:
{"almond", "antique", "aquamarine", "azure", "beige", "bisque", "black", "blanched",
"blue", "blush", "brown", "burlywood", "burnished", "chartreuse", "chiffon", "chocolate",
"coral", "cornflower", "cornsilk", "cream", "cyan", "dark", "deep", "dim", "dodger",
"drab", "firebrick", "floral", "forest", "frosted", "gainsboro", "ghost", "goldenrod",
"green", "grey", "honeydew", "hot", "indian", "ivory", "khaki", "lace", "lavender", "lawn",
"lemon", "light", "lime", "linen", "magenta", "maroon", "medium", "metallic", "midnight", "mint", "misty", "moccasin", "navajo", "navy", "olive", "orange", "orchid", "pale",
"papaya", "peach", "peru", "pink", "plum", "powder", "puff", "purple", "red", "rose",
"rosy", "royal", "saddle", "salmon", "sandy", "seashell", "sienna", "sky", "slate", "smoke",
"snow", "spring", "steel", "tan", "thistle", "tomato", "turquoise", "violet", "wheat",
"white", "yellow"}.
P_MFGR text appended with digit ["Manufacturer",M], where M = random value [1,5].
P_BRAND text appended with digit ["Brand",MN], where N = random value [1,5] and M is defined while
generating P_MFGR.
P_TYPE random string [Types].
P_SIZE random value [1 .. 50].
P_CONTAINER random string [Containers].
P_RETAILPRICE =
(90000 + ((P_PARTKEY/10) modulo 20001 ) + 100 * (P_PARTKEY modulo 1000))/100
P_COMMENT text string [14].
For each row in the PART table, four rows in PARTSUPP table with:
PS_PARTKEY = P_PARTKEY.
PS_SUPPKEY = (PS_PARTKEY + (i * (( S/4 ) + (int)(PS_PARTKEY-1 )/S)))) modulo S + 1 where i is the
ith supplier within [0 .. 3] and S = SF * 10,000.
PS_AVAILQTY random value [1 .. 9,999].
PS_SUPPLYCOST random value [1.00 .. 1,000.00].
PS_COMMENT text string [124].
SF * 150,000 rows in CUSTOMER table with:

C_CUSTKEY unique within [SF * 150,000].
C_NAME text appended with digit ["Customer", C_CUSTKEY].
C_ADDRESS random v-string [25].
C_NATIONKEY random value [0 .. 24].
C_PHONE generated according to Clause 5.2.2.9.
C_ACCTBAL random value [-999.99 .. 9,999.99].
C_MKTSEGMENT random string [Segments].
C_COMMENT text string [73].
For each row in the CUSTOMER table, ten rows in the ORDERS table with:
O_ORDERKEY unique within [SF * 1,500,000 * 4].
Comment: The ORDERS and LINEITEM tables are sparsely populated by generating a key value that causes
the first 8 keys of each 32 to be populated, yielding a 25% use of the key range. Test sponsors must not take
advantage of this aspect of the benchmark. For example, horizontally partitioning the test database onto different devices in order to place unused areas onto separate peripherals is prohibited.
O_CUSTKEY = random value [1 .. (SF * 150,000)].
The generation of this random value must be such that O_CUSTKEY modulo 3 is not zero.
Comment: Orders are not present for all customers. Every third customer (in C_CUSTKEY order) is not
assigned any order.
O_ORDERSTATUS set to the following value:

"F" if all lineitems of this order have L_LINESTATUS set to "F".
"O" if all lineitems of this order have L_LINESTATUS set to "O".
"P" otherwise.
Page 95
O_TOTALPRICE computed as:

sum (L_EXTENDEDPRICE * (1+L_TAX) * (1-L_DISCOUNT)) for all LINEITEM of this order.
O_ORDERDATE uniformly distributed between STARTDATE and (ENDDATE - 151 days).
O_ORDERPRIORITY random string [Priorities].
O_CLERK text appended with digit ["Clerk", C] where C = random value [000000001 .. (SF * 1000)].
O_SHIPPRIORITY set to 0.
O_COMMENT text string [49].
For each row in the ORDERS table, a random number of rows within [1 .. 7] in the LINEITEM table with:
L_ORDERKEY = O_ORDERKEY.
L_PARTKEY random value [1 .. (SF * 200,000)].
L_SUPPKEY = (L_PARTKEY + (i * (( S/4 ) + (int)(L_PARTKEY-1 )/S)))) modulo S + 1
where i is the corresponding supplier within [0 .. 3] and S = SF * 10,000.
L_LINENUMBER unique within [7].
L_QUANTITY random value [1 .. 50].
L_EXTENDEDPRICE = l_quantity * p_retailprice
where p_retailprice is from the part with P_PARTKEY = L_PARTKEY.
L_DISCOUNT random value [0.00 .. 0.10].
L_TAX random value [0.00 .. 0.08].
L_RETURNFLAG set to a value selected as follows:
If L_RECEIPTDATE <= CURRENTDATE
then either "R" or "A" is selected at random
else "N" is selected.
L_LINESTATUS set the following value:
"O" if L_SHIPDATE > CURRENTDATE
"F" otherwise.
L_SHIPDATE = O_ORDERDATE + random value [1 .. 121].
L_COMMITDATE = O_ORDERDATE + random value [30 .. 90].
L_RECEIPTDATE = L_SHIPDATE + random value [1 .. 30].
L_SHIPINSTRUCT random string [Instructions].
L_SHIPMODE random string [Modes].
L_COMMENT text string [27].
25 rows in the NATION table with:

N_NATIONKEY unique value between 0 and 24.
N_NAME string from the following series of (N_NATIONKEY, N_NAME, N_REGIONKEY).
(0, ALGERIA, 0);(1, ARGENTINA, 1);(2, BRAZIL, 1);
(3, CANADA, 1);(4, EGYPT, 4);(5, ETHIOPIA, 0);
(6, FRANCE, 3);(7, GERMANY, 3);(8, INDIA, 2);
(9, INDONESIA, 2);(10, IRAN, 4);(11, IRAQ, 4);
(12, JAPAN, 2);(13, JORDAN, 4);(14, KENYA, 0);
(15, MOROCCO, 0);(16, MOZAMBIQUE, 0);(17, PERU, 1);
(18, CHINA, 2);(19, ROMANIA, 3);(20, SAUDI ARABIA, 4);
(21, VIETNAM, 2);(22, RUSSIA, 3);(23, UNITED KINGDOM, 3);
(24, UNITED STATES, 1)
N_REGIONKEY is taken from the series above.
N_COMMENT text string [95].
5 rows in the REGION table with:

R_REGIONKEY unique value between 0 and 4.
R_NAME string from the following series of (R_REGIONKEY, R_NAME).
(0, AFRICA);(1, AMERICA); (2, ASIA);
(3, EUROPE);(4, MIDDLE EAST)
R_COMMENT text string [95].
Page 96
5.2.4
Refresh Function Data Generation
5.2.4.1
The test database is initially populated with 75% sparse primary keys for the ORDERS and LINEITEM tables (see
Clause 5.2.3) where only the first eight key values of each group of 32 keys are used. Subsequently, the refresh
function RF1 uses the 'holes' in the key ranges for inserting new rows.
5.2.4.2
DBGEN generates refresh data sets for the refresh functions such that:
For the first through the 1,000th execution of RF1 data sets are generated for inserting 0.1% new rows with a
primary key within the second 8 key values of each group of 32 keys;
For the first through the 1,000th execution of RF2 data sets are generated for deleting 0.1% existing rows with
a primary key within the original first 8 key values of each group of 32 keys.
Comment: As a result, after 1,000 executions of RF1/RF2 pairs the test database is still populated with 75% sparse
primary keys, but the second 8 key values of each group of 32 keys are now used.
5.2.4.3
The refresh function data set generation scheme can be repeated until 4000 RF1/RF2 pairs have been executed, at
which point the population of the test database is once again in its initial state.
5.2.5
Database Size
5.2.5.1
Table 3: Estimated Database Size shows the test database size for a scale factor, SF, of 1.
Table 3: Estimated Database Size
Table Name
Cardinality
(in rows)
Length (in bytes)

of Typical2 Row
Typical2 Table
Size (in MB)
SUPPLIER
10,000
159
PART
200,000
155
30
PARTSUPP
800,000
144
110
CUSTOMER
150,000
179
26
ORDERS
1,500,000
104
149
LINEITEM3
6,001,215
112
641
NATION1
25
128
<1
REGION1
124
<1
Total
8,6
61,
24
5
956
Fixed cardinality: does not scale with SF.
2 Typical lengths and sizes given here are examples, not requirements, of what could result from an implementation (sizes do not include storage/access overheads).
3
The cardinality of the LINEITEM table is not a strict multiple of SF since the number of lineitems in an order
Page 97
is chosen at random with an average of four (see Clause 5.2.5.2).

Note: 1 MB is defined to be 220 bytes.
Data types are sized as follows: 4-byte integers, 8-byte decimals, 4-byte dates.
5.2.5.2
Table 4: LINEITEM Cardinality shows the cardinality of the LINEITEM table at all authorized scale factors.
Table 4: LINEITEM Cardinality
Scale Factor (SF)
Cardinality of LINEITEM Table
6001215
10
59986052
30
179998372
100
600037902
300
1799989091
1000
5999989709
3000
18000048306
10000
59999994267
30000
179999978268
100000
599999969200
5.3
Database Load Time
5.3.1
The process of building the test database is known as database load. Database load consists of timed and untimed
components. However, all components must be fully disclosed (see Clause 9.3.5.7).
5.3.2
The total elapsed time to prepare the test database for the execution of the performance test is called the database
load time, and must be reported. This includes all of the elapsed time to create the tables defined in Clause 2.4, load
data, create indices, define and validate constraints, gather statistics for the test database, configure the system
under test as it will be during the performance test, and ensure that the test database meets the ACID requirements
including syncing loaded data on RAID devices and the taking of a backup of the database, when necessary.
5.3.3
The population of the test database, as defined in Clause 5.1, consists of two logical phases:
1. Generation: the process of using DBGEN or a similar program to create data in a format suitable for presentation to the DBMS load facility. The generated data may be stored in memory, or in flat files on tape or disk.
2. Loading: the process of storing the generated data into the database tables.
Generation and loading of the data can be accomplished in two ways:
1. DBGEN (or a similar program) is used to generate flat files stored on disk or tape. The records in these files may
optionally be permuted and relocated to the SUT. After table creation on the SUT, data is loaded from the flat
files into the database. In this case, called load from flat files, only the loading phase contributes to the database
load time.
Page 98
2. DBGEN (or a similar program) is used to generate data that is directly loaded into the database tables using an
"in-line" load facility. In this case, called in-line load, generation and loading occur concurrently and both contribute to the database load time.
5.3.4
The database load time must be measured on the system under test (SUT).
5.3.5
The timing of the database load time begins with the creation of the tables defined in Clause 2.4.
5.3.6
There are five classes of operations which may be excluded from database load time:
Any operation that does not affect the state of the DBMS (e.g., data generation into flat files, relocation of flat
files to the SUT, permutation of data in flat files, operating-system-level disk partitioning or configuration);
Any modification to the state of the DBMS that is not specific to the TPC-H workload (e.g. logical tablespace
creation or database block formatting);
The time required to install or remove physical resources (e.g. processors/cores/threads, memory or disk
drives) on the SUT that are not priced (see Clause 5.3.9);
An optional backup of the test database performed at the test sponsors discretion. However, if a backup is
required to ensure that the ACID properties can be met it must be included in the load time;
Operations that create RAID devices.
Comment: The time required to perform any necessary software reconfiguration (such as DBMS or operating system parameters) must be included in the database load time.
5.3.7
The timing of the database load ends when the database is fully populated and the SUT is configured as it will be
during the performance test.
Comment: The intent of this Clause is that when the timing ends the system under test be capable of executing the
performance test without any further change. The database load may be decomposed into several phases. Database
load time is the sum of the elapsed times of all phases during which activity other than that detailed in Clause 5.3.6
occurred on the SUT. The timing of a load phase completes only when any change to the test database has been
written to durable media (see Clause 4.5.1).
Comment 2: Since the time of the end of the database load is used to seed the random number generator for the
substitution parameter, that time cannot be delayed in any way that would make it predictable to the test sponsor.
5.3.8
5.3.9
The resources used to generate, permute, relocate to the SUT or hold DBGEN data may optionally be distinct from
those used to run the actual benchmark. For example:
For load from flat files, a separate system or a distinct storage subsystem may be used to generate, store and
permute DBGEN data into the flat files used for the database load.
For in-line load, separate and distinct processing elements may be used to generate and permute data and to
deliver it to the DBMS.
Resources used only in the generation phase of the population of the test database must be treated as follows:
For load from flat files,
Any processing element (e.g., processor/core/thread or memory) used exclusively to generate and hold
DBGEN data or relocate it to the SUT prior to the load phase shall not be included in the total priced system
(see Clause 8.1) and must be physically removed from or made inaccessible to the SUT prior to the start of the
load phase;
Page 99
Any storage facility (e.g., disk drive, tape drive or peripheral controller) used exclusively to generate and
deliver data to the SUT during the load phase shall not be included in the total priced system. The test sponsor
must demonstrate to the satisfaction of the auditor that this facility is not being used in the performance tests.
For in-line load,
Any processing element (e.g., processor/core/thread or memory) or storage facility (e.g., disk drive, tape drive
or peripheral controller) used exclusively to generate and deliver DBGEN data to the SUT during the load
phase shall not be included in the total priced system and must be physically removed from or made inaccessible to the SUT prior to the start of the measurement tests.
Comment: The intent is to isolate the cost of resources required to generate data from those required to load data
into the database tables.
5.3.10
An implementation may require additional programs to transfer DBGEN data into the database tables (from either
flat file or in-line load). If non-commercial programs are used for this purpose, their source code must be disclosed.
If commercially available programs are used for this purpose, their invocation and configuration must be disclosed.
Whether or not the software is commercially available, use of the software's functionality's must be limited to:
1. Permutation of the data generated by DBGEN;
2. Delivery of the data generated by DBGEN to the DBMS.
5.3.11
The database load must be implemented using commercially available utilities (invoked at the command level or
through an API) or an SQL programming interface (such as embedded SQL or ODBC).
Page 100
6: PERFORMANCE METRICS AND EXECUTION RULES

6.1
Definition of Terms
6.1.1
Components of the Benchmark
6.1.1.1
The benchmark is defined as the execution of the load test followed by the performance test.
6.1.1.2
The load test begins with the creation of the database tables and includes all activity required to bring the system
under test to the configuration that immediately precedes the beginning of the performance test (see Clause 5.1.3).
The load test may not include the execution of any of the queries in the performance test (see Clause 6.1.2) or any
similar query.
6.1.1.3
The performance test consists of two runs.
6.1.1.4
A run consists of one execution of the Power test described in Clause 6.3.3 followed by one execution of the
Throughput test described in Clause 6.3.4.
6.1.1.5
Run 1 is the first run following the load test (see Clause 6.3.1.4). Run 2 is the run following Run 1.
6.1.1.6
A failed run is defined as a run that did not complete successfully due to unforeseen system failures.
6.1.2
Components of a Run
6.1.2.1
A query is defined as any one of the 22 TPC-H queries specified in Clause 3.
The symbol "Qi ", with i in lowercase and from 1 to 22, represents a given query.
6.1.2.2
A query set is defined as the sequential execution of each and every one of the queries.
6.1.2.3
A query stream is defined as the sequential execution of a single query set submitted by a single emulated user.
The symbol "S", in uppercase, is used to represent the number of query streams used during the throughput
test;
The symbol "s", in lowercase and from 1 to S, is used to represent a given query stream.
6.1.2.4
A refresh stream is defined as the sequential execution of an integral number of pairs of refresh functions submitted from within a batch program.
6.1.2.5
A pair of refresh functions is defined as one of each of the two TPC-H refresh functions specified in Clause 3.
The symbol "RFj ", with j in lowercase and from 1 to 2, represents a given refresh function.
6.1.2.6
A session is defined as the process context capable of supporting the execution of either a query stream or a refresh
stream.
6.2
Configuration Rules
6.2.1
The mechanism used to submit queries and refresh functions to the system under test (SUT) and measure their execution time is called a driver. The driver is a logical entity that can be implemented using one or more physical pro-
Page 101
grams, processes, or systems (see Clause 7.3).

6.2.2
The communication between the driver and the SUT must be limited to one session per query stream or per refresh
stream. These sessions are prohibited from communicating with one another except for the purpose of scheduling
refresh functions (see Clause 6.3.7.8).
6.2.3
All sessions supporting the execution of a query stream must be initialized in exactly the same way. The initialization of the session supporting the execution of the refresh stream may be different than that of the query streams.
All session initialization parameters, settings and commands must be disclosed.
Comment 1: The attributes of the session used in the query stream(s) (see Clause 6.1.2.3) must be the same as the
attributes of the session used by the ACID Query (see Clause 4.1.6.3). Similarly, the attributes of the session used in
the refresh stream (see Clause 6.1.2.4) must be the same as the attributes of the session used by the ACID Transaction (see Clause 4.1.6.3)
Comment 2: The intent of this Clause is to provide the information needed to precisely recreate the execution environment of any given stream prior to the submission of the first query or refresh function.
6.2.4
The driver submits each TPC-H query for execution by the SUT via the session associated with the corresponding
query stream.
6.2.5
In the case of the two refresh functions (RF1 and RF2), the driver is only required to submit the commands necessary to cause the execution of each refresh function.
6.2.6
The driver's submittal to the SUT of the queries in the performance test (see Clause 6.1.2.1) is constrained by the
following restrictions:
It must comply with the query compliance requirements of Clause 3.2;
No part of the interaction between the driver and the SUT can have the purpose of indicating to the DBMS or
operating system an execution strategy or priority that is time dependent or query specific;
Comment: Automatic priority adjustment performed by the operating system is not prohibited, but specifying a
varying priority to the operating system on a query by query basis is prohibited.
6.2.7
The settings of the SUT's components, such as DBMS (including tables and tablespaces) and operating system, are not to be modified on a query by query basis. These parameters have to be set once before any query
or refresh function is run and left in that setting for the duration of the performance test.
The configuration and initialization of the SUT, the database, or the session, including any relevant parameter,
switch or option settings, must be based only on externally documented capabilities of the system that can be reasonably interpreted as useful for an ad-hoc decision support workload. This workload is characterized by:
Sequential scans of large amounts of data;
Aggregation of large amounts of data;
Multi-table joins;
Possibly extensive sorting.
While the configuration and initialization can reflect the general nature of this expected workload, it shall not take
special advantage of the limited functions actually exercised by the benchmark. The queries actually chosen in the
benchmark are merely examples of the types of queries that might be used in such an environment, not necessarily
the actual user queries. Due to this limit in the number and scope of the queries and test environment, TPC-H has
chosen to restrict the use of some database technologies (see Clause 2.5 and Clause 6.2.8). In general, the effect of
Page 102
the configuration on benchmark performance should be representative of its expected effect on the performance of
the class of applications modeled by the benchmark.
Furthermore, the features, switches or parameter settings that comprise the configuration of the operating system,
the DBMS or the session must be such that it would be reasonable to expect a database administrator with the following characteristics be able to decide to use them:
Knowledge of the general characteristics of the workload as defined above;
Knowledge of the logical and physical database layout;
Access to operating system and database documentation;
No knowledge of product internals beyond what is documented externally.
Each feature, switch or parameter setting used in the configuration and initialization of the operating system, the
DBMS or the session must meet the following criteria:
6.2.8
It shall remain in effect without change throughout the performance test;
It shall not make reference to specific tables, indices or queries for the purpose of providing hints to the query
optimizer.
The gathering of statistics is part of the database load (see Clause 5.3) but it also serves as an important configuration vehicle, particularly for the query optimizer. In order to satisfy the requirements of Clause 6.2.7, it is desirable
to collect the same quality of statistics for every column of every table. However, in order to reduce processing
requirements, it is permissible to segment columns into distinct classes and base the level of statistics collection for
a particular column on class membership. Class definitions must rely solely on schema-related attributes of a column and must be applied consistently across all tables. For example:
membership in an index;
leading or other position in an index;
use in a constraint (including a primary or foreign key relationships).
Statistics that operate in sets, such as distribution statistics, should employ a fixed set appropriate to the scale factor
used. Knowledge of the cardinality, values or distribution of a non-key column as specified in Clause 5 cannot be
used to tailor statistics gathering.
6.2.9
Special rules apply to the use of so-called profile-directed optimization (PDO), in which binary executables are
reordered or otherwise optimized to best suit the needs of a particular workload. These rules do not apply to the routine use of PDO by a database vendor in the course of building commercially available and supported database
products; such use is not restricted. Rather, the rules apply to the use of PDO by a test sponsor to optimize executables of a database product for a particular workload. Such optimization is permissible if all of the following conditions are satisfied:
1. The use of PDO or similar procedures by the test sponsor must be disclosed.
2. The procedure and any scripts used to perform the optimization must be disclosed.
3. The procedure used by the test sponsor could reasonably be used by a customer on a shipped database executable.
4. The optimized database executables resulting from the application of the procedure must be supported by the
database software vendor.
5. The workload used to drive the optimization is as described in Clause 6.2.10.
6. The same set of DBMS executables must be used for all phases of the benchmark.
Page 103
6.2.10
If profile-directed optimization is used under the circumstances described in Clause 6.2.9, the workload used to
drive it must be the (possibly repeated) execution of Queries 1,2,4 and 5 in any order, against a TPC-H database of
any desired Scale Factor with default substitution parameters applied.
6.3
Execution Rules
6.3.1
General Rules
6.3.1.1
The driver must submit queries through one or more sessions on the SUT. Each session corresponds to one, and
only one, query stream on the SUT.
6.3.1.2
Parallel activity within the SUT directed toward the execution of a single query (i.e., intra-query parallelism) is not
restricted.
6.3.1.3
To measure the performance of a system using the TPC Benchmark H, the test sponsor will execute runs composed of:
6.3.1.4
6.3.1.5
A power test, to measure the raw query execution power of the system when connected with a single active
user. In this test, a single pair of refresh functions are executed exclusively by a separate refresh stream and
scheduled before and after the execution of the queries (see Clause 6.3.3);
A throughput test, to measure the ability of the system to process the most queries in the least amount of
time. In this test, several pairs of refresh functions are executed exclusively by a separate refresh stream. and
scheduled as defined by the test sponsor.
Comment: The throughput test is where test sponsors can demonstrate the performance of their systems
against a multi-user workload.
The performance test follows the load test. However, any system activity that takes place between the completion of
the load test (see Clause 6.1.1.2) and the beginning of the performance test is limited to that which is not likely to
improve the results of the subsequent performance test. All such activity must be disclosed (see Clause 9.3.6.1).
Examples of acceptable activity include but are not limited to:
Execution of scripts or queries requested by the auditor;
Processing or archiving of files or timing data gathered during the load test;
Configuration of performance monitoring tools;
Execution of simple queries to verify that the database is correctly loaded;
Taking database backups (if not needed to meet the ACID requirements);
Rebooting the SUT or restarting the RDBMS.
The power test and the throughput test must both be executed under the same conditions, using the same hardware
and software configuration and the same data manager and operating system parameters. All such parameters must
be reported.
Comment: The intent of this Clause is to require that both tests (i.e., the power and throughput tests) be run in identical conditions except for the number of query streams and the scheduling of the refresh functions within the
refresh stream.
6.3.1.6
For each query, at least one atomic transaction must be started and completed.
Comment: The intent of this Clause is to specifically prohibit the execution of an entire query stream as a single
transaction.
Page 104
6.3.1.7
Each refresh function must consist of at least one atomic transaction. However, logically consistent portions of the
refresh functions may be implemented as separate transactions as defined in Clause 3.26.
Comment: This intent of this Clause is to specifically prohibit the execution of multiple refresh functions as a single transaction. The splitting of each refresh function into multiple transactions is permitted to encourage "trickle"
updates performed concurrently with one or more query streams in the throughput test.
6.3.2
Run Sequencing
The performance test consists of two runs. If Run 1 is a failed run (see Clause 6.1.1.6) the benchmark must be
restarted with a new load test. If Run 2 is a failed run, it may be restarted without a reload. The reported performance metric must be for the run with the lower TPC-H Composite Query-Per-Hour Performance Metric. The same
set of seed values may be used in the consecutive runs.
The TPC-H metrics reported for a given system must represent a conservative evaluation of the systems level of
performance. Therefore, the reported performance metrics must be for the run with the lower Composite Query-perHour metric
6.3.3
Power Test
6.3.3.1
The power test must be driven by queries submitted by the driver through a single session on the SUT. The session
executes queries one after another. This test is used to measure the raw query execution power of the SUT with a
single query stream. The power test must be executed in parallel with a single refresh stream (see Clause 6.1.2.4).
6.3.3.2
The power test must follow these steps in order:

1. The refresh function RF1 is executed by the refresh stream.
2. The full query set is executed once by the query stream.
3. The refresh function RF2 is executed by the same refresh stream.
6.3.3.3
The timing intervals (see Clause 6.3.7) for each query and for both refresh functions are collected and reported.
6.3.4
Throughput Test
Table 11: Minimum Required Stream Count

SF
S(Streams)
10
30
100
300
1000
3000
10000
30000
10
Page 105
Table 11: Minimum Required Stream Count

SF
100000
6.3.4.1
S(Streams)
11
The throughput test must be driven by queries submitted by the driver through two or more sessions on the SUT.
There must be one session per query stream on the SUT and each stream must execute queries serially (i.e., one
after another). The value of S, the minimum number of query streams, is given in Table 11. The throughput test
must be executed in parallel with a single refresh stream (see Clause 6.1.2.4).
The throughput test must immediately follow one, and only one, power test. No changes to the configuration of the
SUT can be made between the power test and the throughput test (see 5.2.7). Any operations performed on the
SUT between the power and throughput tests must have the following characteristics:
They are related to data collection required for the benchmark or requested by the auditor
They are not likely to improve the performance of the throughput test
6.3.4.2
When measuring and reporting a throughput test, the number, S, of query streams must remain constant during the
whole measurement interval. When results are reported with S query streams, these S streams must be the only ones
executing during the measurement interval (i.e., it is not allowed to execute more than S query streams and report
only the S best ones).
6.3.4.3
For query sequencing purposes (see Clause 6.3.5), each query stream within the throughput test must be assigned a
unique stream identification number ranging from 1 to S, the number of query streams in the test.
6.3.4.4
When measuring and reporting a throughput test, a single refresh stream (see Clause 6.1.2.4) must be executed in
parallel with the S query streams.
6.3.5
Query Sequencing Rules
6.3.5.1
The query sequencing rules apply to each and every query stream, whether part of the power test or part of the
throughput test.
6.3.5.2
Each query set has an ordering number, O(s), based on the identification number, s, of the query stream executing
the set. For example:
The query set within the unique query stream of the power test has the ordering number O(00);
The query set within the first query stream of the throughput test has the ordering number O(01);
The query set within the last of S query streams of the throughput test has the ordering number O(S).
6.3.5.3
The sequencing of query executions is done within a query set. The ordering number, O(s), of a query set determines the order in which queries must be submitted (i.e., sequenced for execution) within that set and is independent of any other query set.
6.3.5.4
The query submission order of an ordering number, O(s), is given in Appendix A by the ordered set with reference
s.
Comment: For tests where the list of ordered sets in Appendix A is exhausted, the last reference in the list must be
followed by the first reference in the list (i.e., wrapping around to s = 00).
Page 106
6.3.6
Measurement Interval
The measurement interval, Ts, for the throughput test is measured in seconds as follows:
It starts either when the first character of the executable query text of the first query of the first query stream is
submitted to the SUT by the driver, or when the first character requesting the execution of the first refresh
function is submitted to the SUT by the driver, whichever happens first;
Comment: In this clause a query stream is said to be first if it starts submitting queries before any other query
streams.
It ends either when the last character of output data from the last query of the last query stream is received by
the driver from the SUT, or when the last transaction of the last refresh function has been completely and successfully committed at the SUT and a success message has been received by the driver from the SUT, whichever happens last.
Comment: In this clause the last query stream is defined to be that query stream whose output data are received last
by the driver.
6.3.7
Timing Intervals
6.3.7.1
Each of the TPC-H queries and refresh functions must be executed in an atomic fashion and timed in seconds.
6.3.7.2
The timing interval, QI(i,s), for the execution of the query, Qi, within the query stream, s, must be measured
between:
The time when the first character of the executable query text is submitted to the SUT by the driver;
The time when the first character of the next executable query text is submitted to the SUT by the driver,
except for the last query of the set for which it is the time when the last character of the query's output data is
received by the driver from the SUT.
Comment: All the operations that are part of the execution of a query (e.g., creation and deletion of a temporary
table or a view) must be included in the timing interval of that query.
6.3.7.3
The timing interval, RI(j,s), for the execution of the refresh function, RFj, within the query stream (for the power
test) or the refresh stream (for the throughput test), where s is 0 for the power test and s is the position of the pair of
refresh functions for the throughput test, must be measured between:
The time when the first character requesting the execution of the refresh function is submitted to the SUT by
the driver;
The last transaction of the refresh function has been completely and successfully committed at the SUT and a
success message has been received by the driver from the SUT.
6.3.7.4
The real-time clock used by the driver to compute the timing intervals must be capable of a resolution of at least
0.01 second.
6.3.7.5
The timing interval of each query and refresh function executed during both tests (i.e., during the power test and the
throughput test) must be individually reported, rounded to the nearest 0.1 second. For example, 23.74 is rounded to
23.7, and 23.75 is rounded to 23.8. Values of less than 0.05 second must be rounded up to 0.1 second to avoid zero
values.
6.3.7.6
The throughput test must include the execution of a single refresh stream. This refresh stream must be used exclusively for the execution of the New Sales refresh function (RF1) and the Old Sales refresh function (RF2).
Page 107
Comment: The purpose of the refresh stream is to simulate a sequence of batched data modifications executing
against the database to bring it up to date with its operational data source.
6.3.7.7
The refresh stream must execute a number of pairs of refresh functions serially (i.e., one RF1 followed by one RF2)
equal to the number of query streams used for the throughput test.
Comment: The purpose of this requirement is to maintain a consistent read/write ratio across a wide range of number of query streams.
6.3.7.8
The scheduling of each refresh function within the refresh stream is left to the test sponsor with the only requirement that a given pair must complete before the next pair can be initiated and that within a pair RF1 must complete
before RF2 can be initiated.
Comment: The intent of this Clause is to allow implementations that execute the refresh functions in parallel with
the ad-hoc queries as well as systems that segregate query executions from database refreshes.
6.3.7.9
The scheduling of individual refresh functions within an instance of RF1 or RF2 is left to the test sponsor as long as
they meet the requirements of Clause 3.26.2 and Clause 3.26.3.
Comment: The intent of this Clause is to allow test sponsors to trickle the scheduling of refresh functions to
maintain a more even refresh load throughout the throughput test.
6.3.7.10
Prior to the execution of the refresh stream the DBGEN data used for RF1 and RF2 may only be generated, permuted and relocated to the SUT. Any other operations on these data, such as data formatting or database activity,
must be included in the execution and the timing of the refresh functions.
6.4
Metrics
TPC-H defines three primary metrics:
The TPC-H Composite Query-per-Hour Metric (QphH@Size) is the performance metric, defined in Clause
6.4.3;
The price-performance metric is the TPC-H Price/Performance ($/QphH) and is defined in Clause 6.4.4;
The Availability Date of the system, defined in Clause 0 of the TPC Pricing Specification Version 1.
No other TPC-H metric exists. However, numerical quantities such as TPC-H Power and TPC-H Throughput
(defined in Clause 5.4.1 and Clause 5.4.2 respectively) and S, the number of query streams in the throughput test,
must be disclosed in the numerical quantities summary (see Clause 9.4.4).
6.4.1
TPC-H Power
6.4.1.1
The results of the power test are used to compute the TPC-H query processing power at the chosen database size. It
is defined as the inverse of the geometric mean of the timing intervals, and must be computed as:
3600 * SF
24
TPC-H Power@Size =
i = 22
j =2
i =1
j =1
QI (i,0) * RI ( j,0)
Where:
QI(i,0) is the timing interval, in seconds, of query Qi within the single query stream of the power test (see
Page 108
Clause 6.3.7)
RI(j,0) is the timing interval, in seconds, of refresh function RFj within the single query stream of the power
test (see Clause 6.3.7)
Size is the database size chosen for the measurement and SF the corresponding scale factor, as defined in
Clause 5.1.3.
Comment: the power numerical quantity is based on a query per hour rate (i.e., factored by 3600).
6.4.1.2
The units of TPC-H Power@Size are Queries per hour * Scale-Factor, reported to one digit after the decimal point,
rounded to the nearest 0.1.
6.4.1.3
The TPC-H Power can also be computed as:
1
3600 * exp
24
TPC-H Power@Size =
6.4.1.4
j =2
i = 22

ln (QI (i,0 )) + ln (RI ( j ,0 )) * SF
j =1
i =1

Where:
ln(x) is the natural logarithm of x
If the ratio between the longest query timing interval and the shortest query timing interval in the power test is
greater than 1000 (i.e., max[QI(i,0)]/min[QI(i,0)] > 1000), then all query timing intervals which are smaller than
max[QI(i,0)]/1000 must be increased to max[QI(i,0)]/1000. The quantity
max[QI(i,0)]/1000 must be treated as a timing interval as specified in Clause 6.3.7.5 for the purposes of computing
the TPC-H Power@Size.
Comment: The adjusted query timings affect only TPC-H Power@Size and no other component of the FDR.
6.4.2
TPC-H Throughput Numerical Quantity
6.4.2.1
The results of the throughput test are used to compute TPC-H Throughput at the chosen database size. It is
defined as the ratio of the total number of queries executed over the length of the measurement interval, and must be
computed as:
TPC-H Throughput@Size = (S*22*3600)/Ts *SF
Where:
Ts is the measurement interval defined in Clause 6.3.6
S is the number of query streams used in the throughput test.
Size is the database size chosen for the measurement and SF the corresponding scale factor, as defined in
Clause 5.1.3.
6.4.2.2
The units of TPC-H Throughput@Size are Queries per hour * Scale-Factor, reported to one digit after the decimal
point, rounded to the nearest 0.1.
6.4.3
The TPC-H Composite Query-Per-Hour Performance Metric
6.4.3.1
The numerical quantities TPC-H Power and TPC-H Throughput are combined to form the TPC-H composite
query-per-hour performance metric which must be computed as:
Page 109
QphH@Size =
Power @ Size * Throughput @ Size
6.4.3.2
The units of QphH@Size are Queries per hour * Scale-Factor, reported to one digit after the decimal point, rounded
to the nearest 0.1.
6.4.4
The TPC-H Price/Performance Metric
6.4.4.1
The TPC-H Price/Performance metric at the chosen database size, TPC-H Price-per-QphH@Size , must be computed using the performance metric QphH@Size as follows:
TPC-H Price-per-QphH@Size = $/QphH@Size
Where:
$ is the total system price in the reported currency. The list of components to be priced is described in
Clause 7.1 of this specification. How to price the components and how to express the total system price
are defined in Clause 7 of the latest revision of the TPC Pricing Specification Version 1.
QphH@Size is the composite query-per-hour performance metric defined in Clause 6.4.3.
Size is the database size chosen for the measurement, as defined in Clause 5.1.3.
6.4.4.2
The units of Price-per-QphH@Size are expressed as in Clause 7 of the latest revision of TPC Pricing Specification
Version 1. In the United States the price performance is expressed as USD per QphH@Size rounded to the highest
cent (e.g., $12.123 must be shown as $12.13USD for price/performance).
6.4.5
Fair Metric Comparison
6.4.5.1
Comparisons of TPC-H benchmark results measured against databases of different sizes are believed to be misleading because database performance and capabilities may not scale up proportionally with an increase in database size
and, similarly, the system price/performance ratio may not scale down with a decrease in database size.
If results measured against different database sizes (i.e., with different scale factors) appear in a printed or electronic communication, then each reference to a result or metric must clearly indicate the database size against which
it was obtained. In particular, all textual references to TPC-H metrics (performance or price/performance) appearing
must be expressed in the form that includes the size of the test database as an integral part of the metrics name; i.e.
including the @size suffix. This applies to metrics quoted in text or tables as well as those used to annotate charts
or graphs. If metrics are presented in graphical form, then the test database size on which metric is based must be
immediately discernible either by appropriate axis labeling or data point labeling.
In addition, the results must be accompanied by a disclaimer stating:
The TPC believes that comparisons of TPC-H results measured against different database sizes are misleading and
discourages such comparisons.
6.4.5.2
Any TPC-H result is comparable to other TPC-H results regardless of the number of query streams used during the
test (as long as the scale factors chosen for their respective test databases were the same).
6.4.6
Required Reporting Components

To be compliant with the TPC-H standard and the TPC's fair use policies, all public references to TPC-H results for
a given configuration must include the following components:
Page 110
The size of the test database, expressed separately or as part of the metric's names (e.g., QphH@10GB);
The TPC-H Performance Metric, QphH@Size;
The TPC-H Price/Performance metric, $/QphH@Size;
The availability date of the priced configuration (see Clause 7 of the TPC Pricing Specification Version 1).
Following are two examples of compliant reporting of TPC-H results:

Example 1: At 10GB the RALF/3000 Server has a TPC-H Composite Query-per-Hour metric of 3010 when run
against a 10GB database yielding a TPC-H Price/Performance of $1,202 per query-per-hour and will be available 1Apr-99.
Example 2: The RALF/3000 Server, which will start shipping on 1-Apr-99, is rated 3,010 QphH@10GB and 1202
$/QphH@10GB.
Page 111
7: SUT AND DRIVER IMPLEMENTATION

7.1
Models of Tested Configurations
7.1.1
The tested and reported configuration(s) is composed of a driver that submits queries to a system under test (SUT).
The SUT executes these queries and replies to the driver. The driver resides on the SUT hardware and software.
7.1.2
Figure 3: Two driver/SUT configurations, a host-based and a client/server configuration illustrates examples of
driver/SUT configurations. The driver is the shaded area. The diagram also depicts the driver/SUT boundary (see
Clause 6.2 and Clause 6.3) where timing intervals are measured.
Figure 3: Two driver/SUT configurations, a host-based and a client/server configuration
DRIVER
Host Systems
DRIVER
Client(s)
Query
Execution
&
Database
Access
*
*
Network
Server(s)
*
Query
Execution
Network
Database
Access
*
Network
Items marked by an * are optional
7.2
System Under Test (SUT) Definition
7.2.1
The SUT consists of:
The host system(s) or server(s) including hardware and software supporting access to the database employed
in the performance test and whose cost and performance are described by the benchmark metrics;
One or more client processing units (e.g., front-end processors/cores/threads, workstations, etc.) that will execute the queries (if used);
The hardware and software components needed to communicate with user interface devices;
The hardware and software components of all networks required to connect and support the SUT components;
Data storage media sufficient to satisfy both the scaling rules in Clause 5 and the ACID properties of Clause 4.
The data storage media must hold all the data described in Clause 5 and be attached to the processing units(s).
7.2.2
All SUT components, as described in Clause 7.2.1, must be commercially available software or hardware products.
7.2.3
An implementation specific layer can be implemented on the SUT. This layer must be logically located between the
driver and the SUT, as depicted by Figure 4: Implementation Specific Layer.
Figure 4: Implementation Specific Layer
Page 112
DRIVER
Exec. Query Text + Row Count
Output Data
Implementation Specific Layer

Commercially Available
Products
(e.g., OS, DBMS, ISQL)
SUT
7.2.4
An implementation specific layer, if present on the SUT, must be minimal, general purpose (i.e., not limited to the
TPC-H queries) and its source code must be disclosed. Furthermore, the functions performed by an implementation
specific layer must be strictly limited to the following:
Database transaction control operations before and after each query execution;
Cursor control and manipulation operations around the executable query text;
Definition of procedures and data structures required to process dynamic SQL, including the communication
of the executable query text to the commercially available layers of the SUT and the reception of the query
output data;
Communication with the commercially available layers of the SUT;
Buffering of the query output data;
Communication with the driver.
The following are examples of functions that the implementation specific layer shall not perform:
Any modification of the executable query text;
Any use of stored procedures to execute the queries;
Any sorting or translation of the query output data;
Any function prohibited by the requirements of Clause 6.2.7.
7.3
Driver Definition
7.3.1
The driver presents the workload to the SUT.
7.3.2
The driver is a logical entity that can be implemented using one or more programs, processes, or systems and performs the functions defined in Clause 7.3.3.
7.3.3
The driver can perform only the following functions:
Generate a unique stream ID, starting with 1 (or 0 for the power test), for each query stream;
Sequence queries for execution by the query streams (see Clause 6.3.5);
Activate, schedule, and/or synchronize the execution of refresh functions in the refresh stream (see Clause
6.3.7.8);
Page 113
Generate the executable query text for each query;
Generate values for the substitution parameters of each query;
Complete the executable query text by replacing the substitution parameters by the values generated for them
and, if needed, replacing the text-tokens by the query stream ID;
Submit each complete executable query text to the SUT for execution, including the number of rows to be
returned when specified by the functional query definition;
Submit each executable refresh function to the SUT for execution;
Receive the output data resulting from each query execution from the SUT;
Measure the execution times of the queries and the refresh functions and compute measurement statistics;
Maintain an audit log of query text and query execution output.
7.3.4
The generation of executable query text used by the driver to submit queries to the SUT does not need to occur on
the SUT and does not have to be included in any timing interval.
7.3.5
The driver shall not perform any function other than those described in Clause 7.3.3. Specifically, the driver shall
not perform any of the following functions:
7.3.6
Performing, activating, or synchronizing any operation other than those mentioned in Clause 7.3.3;
Delaying the execution of any query after the execution of the previous query other than for delays necessary
to process the functions described in Clause 7.3.3. This delay must be reported and can not exceed half a second between any two consecutive queries of the same query stream;
Modifying the compliant executable query text prior to its submission to the SUT;
Embedding the executable query text within a stored procedure definition or an application program;
Submitting to the SUT the values generated for the substitution parameters of a query other than as part of the
executable query text submitted;
Submitting to the SUT any data other than the instructions to execute the refresh functions, the compliant executable query text and, when specified by the functional query definition, the number of rows to be returned;
Artificially extending the execution time of any query.
The driver is not required to be priced.
Page 114
8: PRICING
This section defines the components, functional requirements of what is priced, and what substitutions are allowed.
Rules for pricing the Priced Configuration and associated software and maintenance are included in the current
revision of the TPC Pricing Specification Version 1 located at www.tpc.org.
8.1
Priced System
The system to be priced shall include the hardware and software components present in the System Under Test
(SUT), a communication interface that can support user interface devices, additional operational components configured on the test system, and maintenance on all of the above
8.1.1
System Under Test

Calculation of the priced system consists of:
Price of the SUT as tested and defined in Clause 7;
Price of a communication interface capable of supporting the required number of user interface devices
defined in Clause 8.1.2.1;
Price of on-line storage for the database as described in Clause 8.1.3 and storage for all software included in
the priced configuration;
Price of additional products (software or hardware) required for customary operation, administration and
maintenance of the SUT for a period of 3 years
Price of all products required to create, execute, administer, and maintain the executable query texts or necessary to create and populate the test database.
Specifically excluded from the priced system calculation are:
End-user communication devices and related cables, connectors, and concentrators;
Equipment and tools used exclusively in the production of the full disclosure report;
Equipment and tools used exclusively for the execution of the DBGEN or QGEN (see Clause 3.1.4 and Clause
5.2.1) programs.
8.1.2
User Interface Devices and Communications
8.1.2.1
The priced system must include the hardware and software components of a communication interface capable of
supporting a number of user interface devices (e.g., terminals, workstations, PCs, etc.) at least equal to 10 times the
number of query streams used for the throughput test (see Clause 6.3.4).
Comment: Test sponsors are encouraged to configure the SUT with a general-purpose communication interface
capable of supporting a large number of user interface devices.
8.1.2.2
Only the interface is to be priced. Not to be included in the priced system are the user interface devices themselves
and the cables, connectors and concentrators used to connect the user interface devices to the SUT. For example, in
a configuration that includes an Ethernet interface to communicate with PCs, the Ethernet card and supporting software must be priced, but not the Ethernet cables and the PCs.
Comment: Active components (e.g., workstations, PCs, concentrators, etc.) can only be excluded from the priced
system under the assumption that their role is strictly limited to submitting executable query text and receiving output data and that they do not participate in the query execution. All query processing performed by the tested conTPC Benchmark H Standard Specification Revision 2.8.0
Page 115
figuration is considered part of the performance test and can only be done by components that are included in the
priced system.
8.1.2.3
The communication interface used must be an industry standard interface, such as Ethernet, Token Ring, or RS232.
8.1.2.4
The following diagram illustrates the boundary between what is priced (on the right) and what is not (on the left):
Figure 5: The Pricing Boundary
Driver
SUT
(Implementation Specific Layer)
User Interface
Device(s)
Network
Commercially Available
Products
(e.g., OS, DBMS, ISQL)
Pricing Boundary
8.1.3
Database Storage and Recovery Log
8.1.3.1
Recovery data must be maintained for at least the duration of the run used to compute the published performance
metrics.(see Clause 6.1.1.4).
Roll-back recovery data must be either in memory or in on-line storage at least until all transactions dependent on it
are committed. Roll-forward recovery data may be stored on an off-line device provided that:
The process that stores the roll-forward data is active during the measurement interval;
The roll-forward data that is stored off-line during the measurement interval must be at least as great as the
roll-forward recovery data that is generated during the period (i.e., the data may be first created in on-line storage and then moved to off-line storage, but the creation and the movement of the data must be in steady state);
All ACID properties must be retained.
Comment: Storage is considered on-line if any record can be accessed randomly and updated within 1 second even
if this access time requires the creation of a logical access path not present in the tested database. For example, a
disk-based sequential file might require the creation of an index to satisfy the access time requirement. On-line storage may include magnetic disks, optical disks, or any combination of these, provided that the above mentioned
access criteria are met.
8.1.3.2
While the benchmark requires the configuration of storage sufficient to hold the requisite recovery data as specified
in Clause 8.1.3.1, it does not explicitly require the demonstration of rollforward recovery except as required by the
ACID tests (See Clause 4.5).
8.1.3.3
This clause has been left intentially blank.
8.1.3.4
The storage that is required to be priced includes:
storage required to execute the benchmark;
Page 116
storage to hold recovery data (see Clause 8.1.3);
storage and media needed to assure that the test database meets the ACID requirements defined in Clause 4.
8.1.3.5
All storage required for the priced system must be present on the tested system.
8.1.4
Additional Operational Components
8.1.4.1
Additional products that might be included on a customer installed configuration, such as operator consoles and
magnetic tape drives, are also to be included in the priced system if explicitly required for the operation, administration, or maintenance, of the priced system.
8.1.4.2
Copies of the software, on appropriate media, and a software load device, if required for initial load or maintenance
updates, must be included.
8.1.4.3
The price of an Uninterruptible Power Supply, if specifically contributing to a durability solution, must be included
(see Clause 4.5).
8.1.4.4
The price of all cables used to connect components of the system (except as noted in Clause 8.1.2.2) must be
included.
8.1.5
Software
8.1.5.1
All software licenses must be priced for a number of users at least equal to 10 times the number of query streams
used for the multi-stream throughput test (see Clause 5.3.4). Any usage pricing for this number of users must be
based on the pricing policy of the company supplying the priced component.
8.2
Allowable Substitutions
8.2.1
Substitution is defined as a deliberate act to replace components of the Priced Configuration by the test sponsor as a
result of failing the availability requirements of the TPC Pricing Specification or when the part number for a component changes.
Comment 1: Corrections or "fixes" to components of the Priced Configuration are often required during the life of
products. These changes are not considered Substitutions so long as the part number of the priced component does
not change. Suppliers of hardware and software may update the components of the Priced Configuration, but these
updates must not impact the reported performance metric or numerical quantities. The following are not considered
substitutions:
software patches to resolve a security vulnerability
silicon revision to correct errors
new supplier of functionally equivalent components (i.e. memory chips, disk drives, ...)
Durable Medium is defined as a data storage medium that is inherently non-volatile such as a magnetic disk or tape.
8.2.2
Some hardware components of the Priced Configuration may be substituted after the test sponsor has demonstrated
to the auditor's satisfaction that the substituting components do not negatively impact the reported performance
metric or numerical quantities. All substitutions must be reported in the FDR and noted in the auditor's attestation
letter. The following hardware components may be substituted:
Durable Medium, Disk Enclosure, external storage controllers
Page 117
Network interface cards
Routers, Bridges, Repeaters, Switches
Cables
Page 118
9: FULL DISCLOSURE
Rules for reporting Pricing information are included in the current revision of the TPC Pricing Specification Version 1 located at www.tpc.org.
9.1
Reporting Requirements
9.1.1
A Full Disclosure Report (FDR) and Executive Summary are required.
9.1.2
The intent of this disclosure is to simplify comparison between results and for a customer to be able to replicate the
results of this benchmark given appropriate documentation and products.
9.2
Format Guidelines
9.2.1
While established practice or practical limitations may cause a particular benchmark disclosure to differ from the
examples provided in various small ways, every effort should be made to conform to the format guidelines. The
intent is to make it as easy as possible for a reviewer to read, compare and evaluate material in different benchmark
disclosures.
9.2.2
All sections of the report, including appendices, must be printed using font sizes of a minimum of 8 points.
9.2.3
The Executive Summary must be included near the beginning of the full disclosure report.
9.3
Full Disclosure Report Contents

The FDR should be sufficient to allow an interested reader to evaluate and, if necessary, recreate an implementation
of TPC-H. If any sections in the FDR refer to another section of the report (e.g., an appendix), the names of the referenced scripts/programs must be clearly labeled in each section.
Comment: Since the building of a database may consist of a set of scripts and corresponding input files, it is important to disclose and clearly identify, by name, scripts and input files in the FDR.
The order and titles of sections in the test sponsor's full disclosure report must correspond with the order and titles
of sections from the TPC-H standard specification (i.e., this document).
9.3.1
General Items
9.3.1.1
A statement identifying the benchmark sponsor(s) and other participating companies must be provided.
9.3.1.2
Settings must be provided for all customer-tunable parameters and options that have been changed from the defaults
found in actual products, including but not limited to:
Database tuning options;
Optimizer/Query execution options;
Query processing tool/language configuration parameters;
Recovery/commit options;
Consistency/locking options;
Operating system and configuration parameters;
Configuration parameters and options for any other software component incorporated into the pricing struc-
Page 119
ture;
Compiler optimization options.
Comment 1: In the event that some parameters and options are set multiple times, it must be easily discernible by
an interested reader when the parameter or option was modified and what new value it received each time.
Comment 2: This requirement can be satisfied by providing a full list of all parameters and options, as long as all
those that have been modified from their default values have been clearly identified and these parameters and
options are only set once.
9.3.1.3
Explicit response to individual disclosure requirements specified in the body of earlier sections of this document
must be provided.
9.3.1.4
Diagrams of both measured and priced configurations must be provided, accompanied by a description of the differences. This includes, but is not limited to:
Total number of nodes used, total number and type of processors used/total number of cores used/total number
of threads used (including sizes of L2 and L3 caches);
Size of allocated memory, and any specific mapping/partitioning of memory unique to the test;
Number and type of disk units (and controllers, if applicable);
Number of channels or bus connections to disk units, including their protocol type;
Number of LAN (e.g., Ethernet) connections, including routers, workstations, terminals, etc., that were physically used in the test or are incorporated into the pricing structure;
Type and the run-time execution location of software components (e.g., DBMS, query processing tools/languages, middleware components, software drivers, etc.).
The following sample diagram illustrates a measured benchmark configuration using Ethernet, an external driver,
and four processors each with two cores and four threads per node in the SUT. Note that this diagram does not
depict or imply any optimal configuration for the TPC-H benchmark measurement.
Figure 1: Sample Configuration Diagram (the front system box describes one node)
Page 120
LAN: Ethernet using NETplus routers

Total number of nodes used/total number of processors used/total number of cores used/total number of threads
used: 4/16/32/64 x a243DX 3GHz with 4 MByte Second Level Cache
4 gigabyte of main memory
16 x SCSI-2 Fast Controllers
Disk: 96 x 2.1 gigabyte SCSI-2 drives
Comment: Detailed diagrams for system configurations and architectures can vary widely, and it is impossible to
provide exact guidelines suitable for all implementations. The intent here is to describe the system components and
connections in sufficient detail to allow independent reconstruction of the measurement environment. This example diagram shows homogeneous nodes. This does not preclude tests sponsors from using heterogeneous nodes as
long as the system diagram reflects the correct system configuration.
9.3.2
Clause 1 - Logical Database Design Related Items
9.3.2.1
Listings must be provided for all table definition statements and all other statements used to set-up the test and qualification databases.
9.3.2.2
The physical organization of tables and indices within the test and qualification databases must be disclosed. If the
column ordering of any table is different from that specified in Clause 1.4, it must be noted.
Comment: The concept of physical organization includes, but is not limited to: record clustering (i.e., rows from
different logical tables are co-located on the same physical data page), index clustering (i.e., rows and leaf nodes of
an index to these rows are co-located on the same physical data page), and partial fill-factors (i.e., physical data
pages are left partially empty even though additional rows are available to fill them).
9.3.2.3
Horizontal partitioning of tables and rows in the test and qualification databases (see Clause ) must be disclosed.
9.3.2.4
Any replication of physical objects must be disclosed and must conform to the requirements of Clause 1.5.10.
9.3.3
Clause 2 - Query and Refresh function-Related Items
9.3.3.1
The query language used to implement the queries must be identified (e.g., RALF/SQL-Plus).
9.3.3.2
The method of verification for the random number generation must be described unless the supplied DBGEN and
QGEN were used.
9.3.3.3
The method used to generate values for substitution parameters must be disclosed. If QGEN is not used for this purpose, then the source code of any non-commercial tool used must be disclosed. If QGEN is used, the version number, release number, modification number and patch level of QGEN must be disclosed.
9.3.3.4
The executable query text used for query validation must be disclosed along with the corresponding output data
generated during the execution of the query text against the qualification database. If minor modifications (see
Clause 2.2.3) have been applied to any functional query definitions or approved variants in order to obtain executable query text, these modifications must be disclosed and justified. The justification for a particular minor query
modification can apply collectively to all queries for which it has been used. The output data for the power and
throughput tests must be made available electronically upon request.
Comment: For query output of more than 10 rows, only the first 10 need to be disclosed in the FDR. The remaining
rows must be made available upon request.
9.3.3.5
All the query substitution parameters used during the performance test must be disclosed in tabular format, along
with the seeds used to generate these parameters.
Page 121
9.3.3.6
The isolation level used to run the queries must be disclosed. If the isolation level does not map closely to one of the
isolation levels defined in Clause 3.4, additional descriptive detail must be provided.
9.3.3.7
The details of how the refresh functions were implemented must be disclosed (including source code of any noncommercial program used).
9.3.4
Clause 3 - Database System Properties Related Items
9.3.4.1
The results of the ACID tests must be disclosed along with a description of how the ACID requirements were met.
This includes disclosure of the code written to implement the ACID Transaction and Query.
9.3.5
Clause 4 - Scaling and Database Population Related Items
9.3.5.1
The cardinality (e.g., the number of rows) of each table of the test database, as it existed at the completion of the
database load (see Clause 4.2.5), must be disclosed.
9.3.5.2
The distribution of tables and logs across all media must be explicitly described using a format similar to that shown
in the following example for both the tested and priced systems.
Comment: Detailed diagrams for layout of database tables on disks can widely vary, and it is difficult to provide
exact guidelines suitable for all implementations. The intent is to provide sufficient detail to allow independent
reconstruction of the test database. The table below is an example of database layout descriptions and is not
intended to describe any optimal layout for the TPC-H database.
Table 12: Sample Database Layout Description
Controller
Disk Drive
Description of Content
40A
Operating system, root
System page and swap
Physical log
100% of PART and SUPPLIER tables
33% of CUSTOMER, ORDERS and LINEITEM tables
100% of PARTSUPP, NATION and REGION tables
40B
9.3.5.3
The mapping of database partitions/replications must be explicitly described.

Comment: The intent is to provide sufficient detail about partitioning and replication to allow independent reconstruction of the test database.
9.3.5.4
Implementations may use some form of RAID. The RAID level used must be disclosed for each device. If RAID is
used in an implementation, the logical intent of its use must be disclosed. Three levels of usage are defined:
Base tables only: In this case only the Base Tables (see Clause 1.2) are protected by any form of RAID;
Base tables and auxiliary data structures: in addition to the protection of the base tables, implementations in
this class must also employ RAID to protect all auxiliary data structures;
Page 122
Everything: implementations in this usage category must employ RAID to protect all database storage, including temporary or scratch space in addition to the base tables and auxiliary data structures.
9.3.5.5
The version number, release number, modification number, and patch level of DBGEN must be disclosed. Any
modifications to the DBGEN (see Clause 4.2.1) source code (see Appendix B) must be disclosed. In the event that a
program other than DBGEN was used to populate the database, it must be disclosed in its entirety.
9.3.5.6
The database load time for the test database (see Clause 4.3) must be disclosed.
9.3.5.7
The data storage ratio must be disclosed. It is computed by dividing the total data storage of the priced configuration (expressed in GB) by the size chosen for the test database as defined in Clause 4.1.3.1. The ratio must be
reported to the nearest 1/100th, rounded up. For example, a system configured with 96 disks of 2.1 GB capacity for
a 100GB test database has a data storage ratio of 2.02.
Comment: For the reporting of configured disk capacity, gigabyte (GB) is defined to be 2^30 bytes. Since disk
manufacturers typically report disk size using base ten (i.e., GB = 10^9), it may be necessary to convert the advertised size from base ten to base two.
9.3.5.8
The details of the database load must be disclosed, including a block diagram illustrating the overall process. Disclosure of the load procedure includes all steps, scripts, input and configuration files required to completely reproduce the test and qualification databases.
9.3.5.9
Any differences between the configuration of the qualification database and the test database must be disclosed.
9.3.6
Clause 5 - Performance Metrics and Execution Rules Related Items
9.3.6.1
Any system activity on the SUT that takes place between the conclusion of the load test and the beginning of the
performance test must be fully disclosed including listings of scripts or command logs.
9.3.6.2
The details of the steps followed to implement the power test (e.g., system boot, database restart, etc.) must be disclosed.
9.3.6.3
The timing intervals (see Clause 5.3.7) for each query and for both refresh functions must be reported for the power
test.
9.3.6.4
The number of query streams used for the throughput test must be disclosed.
9.3.6.5
The start time and finish time for each query stream must be reported for the throughput test.
9.3.6.6
The total elapsed time of the measurement interval (see Clause 5.3.6) must be reported for the throughput test.
9.3.6.7
The start time and finish time for each refresh function in the refresh stream must be reported for the throughput
test.
9.3.6.8
The timing intervals (see Clause 5.3.7) for each query of each stream and for each refresh function must be reported
for the throughput test.
9.3.6.9
The computed performance metric, related numerical quantities and the price/performance metric must be reported.
9.3.6.10
The performance metric (QphH@Size) and the numerical quantities (TPC-H Power@Size and TPC-H Throughput@Size) from both of the runs must be disclosed (see Clause 5.4.1).
Page 123
9.3.6.11
Any activity on the SUT that takes place between the conclusion of Run1 and the beginning of Run2 must be fully
disclosed including listings of scripts or command logs along with any system reboots or database restarts.
9.3.7
Clause 6 - SUT and Driver Implementation Related Items
9.3.7.1
A detailed textual description of how the driver performs its functions, how its various components interact and any
product functionalities or environmental settings on which it relies must be provided. All related source code,
scripts and configuration files must be disclosed. The information provided should be sufficient for an independent
reconstruction of the driver.
9.3.7.2
If an implementation specific layer is used, then a detailed description of how it performs its functions, how its various components interact and any product functionalities or environmental setting on which it relies must be provided. All related source code, scripts and configuration files must be disclosed. The information provided should
be sufficient for an independent reconstruction of the implementation specific layer.
9.3.7.3
If profile-directed optimization as described in Clause 5.2.9 is used, such use must be disclosed. In particular, the
procedure and any scripts used to perform the optimization must be disclosed.
9.3.8
Clause 9 - Audit Related Items
9.3.8.1
The auditor's agency name, address, phone number, and attestation letter with a brief audit summary report indicating compliance must be included in the full disclosure report. A statement should be included specifying whom to
contact in order to obtain further information regarding the audit process.
9.4
Executive Summary
The executive summary is meant to be a high level overview of a TPC-H implementation. It should provide the
salient characteristics of a benchmark execution (metrics, configuration, pricing, etc.) without the exhaustive detail
found in the FDR. The executive summary has three components:
9.4.1
Implementation Overview
Pricing Spreadsheet
Numerical Quantities
Page Layout
Each component of the executive summary should appear on a page by itself. Each page should use a standard
header and format, including
1/2 inch margins, top and bottom;
3/4 inch left margin, 1/2 inch right margin;
2 pt. frame around the body of the page. All interior lines should be 1 pt.;
Sponsor identification and System identification, each set apart by a 1 pt. rule, in 16-20 pt. Times Bold font;
Benchmark name(i.e., TPC-H), revision using three tier versioning(e.g., 1.2.3) and report date, separated from
other header items and each other by a 1 pt. Rule, in 9-12 pt. Times font.
Comment 1: It is permissible to use or include company logos when identifying the sponsor.
Comment 2: The report date must be disclosed with a precision of 1 day. The precise format is left to the test sponsor.
Page 124
Note: Appendix E contains a sample executive summary. It is meant to help clarify the requirements in section 8.4
and is provided solely as an example.
9.4.2
Implementation Overview
The implementation overview page contains six sets of data, each laid out across the page as a sequence of boxes
using 1 pt. rule, with a title above the required quantity. Both titles and quantities should use a 9-12 pt. Times font
unless otherwise noted.
9.4.2.1
The first section contains the results that were obtained from the reported run of the Performance test.
Table 13: Implementation Overview Information
Title
Quantity
Precision
Units
Font
Total System Cost
3 yr. Cost of ownership (see

Clause 7)
$1
16-20 pt. Bold
TPC-H Composite Query

per Hour Metric
QphH (see Clause 5.4.3)
0.1
QphH@nGB
16-20 pt. Bold
Price/Performance
$/QphH (see Clause 5.4.4)
$/QphH@nGB
16-20 pt. Bold
9.4.2.2
The next section details the system configuration

Table 14: System Configuration Information
Title
Quantity
Precision
Units
Font
Database Size
Raw data size of test database

(see Clause 4.1.3 and Clause
8.3.5.7)
GB
(see Clause 8.3.5.7)
9-12 pt. Times
DBMS Manager
Brand, Software Version

DBMS used
of
9-12 pt. Times
Operating System
Brand, Software Version of

OS used
9-12 pt. Times
Other Software
Brand, Software Version of

other software components
9-12 pt. Times
System Availability Date
The Availability Date of the

system, defined in Clause 0 of
the TPC Pricing Specification
Version 1.
1 day
9-12 pt. Times
Comment: The Software Version must uniquely identify the orderable software product referenced in the Priced
Configuration (e.g., RALF/2000 4.2.1)
9.4.2.3
This section is the largest in the implementation overview, and contains a graphic representation of the reported
query times. Each query and refresh function executed during the benchmark should be listed in the graph, with any
query variants clearly identified. In addition:
All labels and scales must use a 10 point Courier font, except for the legend and the graph title which must use
a Times font;
All line sizes must be 1 point;

Page 125
The legend must be reproduced as depicted in the example, and must be placed where needed to avoid overlapping any portion of the graph;
The query time axis must labeled with no more than 8 values, including the zero origin;
Each pair of bars must be separated by a gap of 50% of the bar's width;
A zero-based linear scale must be used for the query times;
The upper bound of the time scale must be no greater than 120% of the longest query timing interval;
The bars used for the power test must be sized based on the measured (i.e., without the adjustment defined in
Clause 5.4.1.4) query timing intervals of the power test, and must be solid white;
The bars used for the throughput test must be sized based on the arithmetic mean by query type of the measured query timing intervals of the throughput test, and must be solid black;
The geometric mean of the power test components must be computed using unadjusted timings of queries and
refresh functions and must be placed on the graph as a dashed line labeled on top with its value. It must be
expressed using the same format and precision as TPC-H Power specified in Clause 5;
The arithmetic mean of the throughput test must be calculated using unadjusted timings with the following
computation:
22
QI ( i, s )
i = 1s = 1
------------------------------------( S 22 )
where QI(i,s) is defined in Clause 5.3.7.2, and S is defined in Clause 5.1.2.3;
9.4.2.4
A solid line representing the mean must be placed on the graph intersecting only the queries and must be
labeled on top with its value. The arithmetic mean of the throughput test must be expressed with the same format and precision as TPC-H Throughput specified in Clause 5;
All query numbers must be followed by a variant letter when a variant was used in the tests.
This section contains the database load and sizing information

Table 15: Database Load and Sizing Information
Title
Quantity
Precision
Units
Font
Database Load Time
Load Time (see Clause 4.3)
1 sec.
hh:mm:ss
9-12 pt. Times
Total Disk/Database Size
Data Storage Ratio (see Clause

8.3.5.7)
0.01
Load includes backup
Y/N (see Clause 4.3.6)
N/A
N/A
9-12 pt. Times
RAID (Base tables only)
Y/N (see Clause 8.3.5.4)
N/A
N/A
9-12 pt. Times
RAID (Base tables and

auxiliary data structures)
N/A
N/A
9-12 pt. Times
RAID (Everything)
N/A
N/A
9-12 pt. Times
9.4.2.5
9-12 pt. Times
The next section of the Implementation Overview should contain a synopsis of the SUT's major system components, including
Page 126
total number of nodes used/total number of processors used with their types and speeds in GHz/ total number
of cores used/total number of threads used;
Main and cache memory sizes;
Network and I/O connectivity;
Disk quantity and geometry.
If the implementation used a two-tier architecture, front-end and back-end systems should be detailed separately.
9.4.2.6
The final section of the implementation Overview should contain a note stating:
Database Size includes only raw data (e.g., no temp, index, redundant storage space, etc.).
9.4.3
Pricing Spreadsheet
The major categories in the Price Spreadsheet, as appropriate, are:
Server Hardware
Server Storage
Server Software
Discounts (may optionally be included with above major category subtotal calculations)t.
9.4.4
Numerical Quantities Summary

The Numerical Quantities Summary page contains three sets of data, presented in tabular form, detailing the execution timings for the reported execution of the performance test. Each set of data should be headed by its given title
and clearly separated from the other tables.
9.4.4.1
The first section contains measurement results from the benchmark execution.
Se ction Title: Measurement Results
Item Title
Precision
Database Scale Factor
Total Data Storage/Database Size
0.01
Start of Database Load
yyyy-mm-dd hh:mm:ss
End of Database Load
yyyy-mm-dd hh:mm:ss
Database Load Time
hh:mm:ss
Query Streams for Throughput Test
TPC-H Power
0.1
TPC-H Throughput
0.1
TPC-H Composite Query-per-Hour Metric (QphH@Size)
0.1
Notes
Page 127
Item Title
Precision
Notes
Total System Price Over 3 Years
$1
(1)
TPC-H Price Performance Metric ($/QphH@Size)
$0.01
(1)
(1) depending on the currency used for publication this sign has to be exchanged with the ISO currency symbol
9.4.4.2
The second section contains query and query stream timing information.
Section Title: Measurement Intervals
Item Title
Precision
Measurement Interval in Throughput Test (Ts)
1 second
Duration of Stream Execution
Notes
(1)
Stream
Seed
Start Date/Time
mm/dd/yy hh:mm:ss
End Date/Time
mm/dd/yy hh:mm:ss
Total Time
hh:mm:ss
Refresh Start Date/Time
mm/dd/yy hh:mm:ss
Refresh End Date/Time
mm/dd/yy hh:mm:ss
(1) The remaining items in this section should be reported as a sub-table, with one entry for each stream executed
during the performance test.
9.4.4.3
The final section, titled Timing Intervals (in Sec.) contains individual query and refresh function timings. The data
should be presented as a table with one entry for each query stream executed during the Performance Test. For each
stream entry, the total elapsed time for each query in the stream and for its associated refresh functions should be
reported separately to a resolution of 0.1 seconds. In addition, the minimum, maximum and average execution time
for each query and refresh function must be reported to a resolution of 0.1 seconds.
9.5
Availability of the Full Disclosure Report
9.5.1
The full disclosure report must be readily available to the public at a reasonable charge, similar to charges for comparable documents by that test sponsor. The report must be made available when results are made public. In order
to use the phrase TPC Benchmark H, the full disclosure report must have been submitted in hard copy and electronically to the TPC using the procedure described in the TPC Policies and Guidelines document.
9.5.2
The official full disclosure report must be available in English but may be translated to additional languages.
9.6
Revisions to the Full Disclosure Report

Revisions to the full disclosure documentation shall be handled as follows:
Page 128
9.6.1
Substitutions will be open to challenge for a 60 day period. No other portion of the FDR is challengable.
9.6.2
During the normal product life cycle, problems will be uncovered that require changes, sometimes referred to as
ECOs, FCOs, patches, updates, etc. When the cumulative result of applied changes causes the QphH/QphR rating
of the system to decrease by more than 2% from the initially reported QphH/QphR, then the test sponsor is required
to re-validate the benchmark results. The complete revision history is maintained following the query timing interval section showing the revision date and description.
9.6.3
Full disclosure report revisions may be required for other reasons according to TPC policies (see TPC Policy Document)
Page 129
10: AUDIT
Rules for auditing Pricing information are included in the current revision of the TPC Pricing Specification Version
1 located at www.tpc.org.
10.1
General Rules
10.1.1
An independent audit of the benchmark results by a TPC certified auditor is required. The term independent is
defined as the outcome of the benchmark carries no financial benefit to the auditing agency other than fees earned
directly related to the audit. In addition, the auditing agency cannot have supplied any performance consulting
under contract for the benchmark.
In addition, the following conditions must be met:
a)
The auditing agency cannot be financially related to the sponsor. For example, the auditing agency is financially related if it is a dependent division of the sponsor, the majority of its stock is owned by the sponsor, etc.
b)
The auditing agency cannot be financially related to any one of the suppliers of the measured/priced configuration, e.g., the DBMS supplier, the disk supplier, etc.
10.1.2
The auditor's attestation letter is to be made readily available to the public as part of the full disclosure report. A
detailed report from the auditor is not required.
10.1.3
TPC-H results can be used as the basis for new TPC-H results if and only if:
a)
The auditor ensures that the hardware and software products are the same as those used in the prior result;
b)
The auditor reviews the FDR of the new results and ensures that they match what is contained in the original
sponsor's FDR;
c)
The auditor can attest to the validity of the pricing used in the new FDR.
Comment 1: The intent of this clause is to allow a reseller of equipment from a given supplier to publish under the
re-seller's name a TPC-H result already published by the supplier.
Comment 2: In the event that all conditions listed in Clause 9.1.3 are met, the auditor is not required to follow the
remaining auditor's check list items from Clause 10.2.
10.1.4
Ensure that any auxiliary data structures satisfy the requirements of Clause 2.5.10.
10.1.5
In the event that a remote audit procedure is used in the context of a change-based audit, a remote connection to the
SUT must be available for the auditor to verify selected audit items from Clause 10.2.
10.2
Auditor's Check List
10.2.1
Clause 1 Related Items
10.2.1.1
Verify that the data types used for each column are conformant. For example, verify that decimal columns can be
incremented by 0.01 from -9,999,999,999.99.
10.2.1.2
Verify that the tables have the required list of columns.
10.2.1.3
Verify that the implementation rules are met by the test database.
Page 130
10.2.1.4
Verify that the test database meets the data access transparency requirements.
10.2.1.5
Verify that conforming arbitrary data values can be inserted into any of the tables. Examples of verification tests
include:
10.2.1.6
Inserting a row that is a complete duplicate of an existing row except for a distinct primary key;
Inserting a row with column values within the domain of the data type and check constraints but beyond the
range of existing values.
Verify that the set of auxiliary data structures (as defined in Clause 2.5.7) that exist at the end of the load test are the
same as those which exist at the end of the performance test. A similar check may be performed at any point during
the performance test at the discretion of the auditor.
Comment: The purpose of this check is to verify that no auxiliary data structures automatically generated during
the performance test may be accessed by more than one query execution.
10.2.2
10.2.2.1
Verify that the basis for the SQL used for each query is either the functional query definition or an approved variant.
10.2.2.2
Verify that any deviation in the SQL from either the functional query definition or an approved variant is compliant
with the specified minor query modifications. Verify that minor query modifications have been applied consistently
to the set of functional query definitions or approved variants used.
10.2.2.3
Verify that the executable query text produces the required output when executed against the qualification database
using the validation values for substitution parameters.
10.2.2.4
Note the method used to generate the values for substitution parameters (i.e., QGEN, modified version of QGEN,
other method). If QGEN was used, note the version number, release number, modification number and patch level
of QGEN. Verify that the version and release numbers match the benchmark specification.
10.2.2.5
Verify that the generated substitution parameters are reasonably diverse among the streams.
10.2.2.6
Verify that no aspect of the system under test, except for the database size, has changed between the demonstration
of compliance against the qualification database and the execution of the reported measurements.
10.2.2.7
Verify that the refresh functions are implemented according to their definition.
10.2.2.8
Verify that the transaction requirements are met by the implementation of the refresh functions.
10.2.2.9
Note the method used to execute database maintenance operations
10.2.3
10.2.3.1
Verify that the required ACID properties are supported by the system under test as configured for the execution of
the reported measurements.
10.2.3.2
If one or more of the ACID tests defined in Clause 4 were not executed, note the rationale for waiving such demonstration of support of the related ACID property.
Page 131
10.2.4
10.2.4.1
Verify that the qualification database is properly scaled and populated.
10.2.4.2
Verify that the test database is properly scaled.
10.2.4.3
Verify that the rows in the loaded database after the performance test are correct by comparing some small number
of rows extracted at random from any two files of the corresponding Base, Insert and Delete reference data set files
for each table and the corresponding rows of the database.
10.2.4.4
Verify that the dbgen (using the command lines provided in Appendix F) used in the benchmark generates a data set
which matches the reference data set provided in Appendix F corresponding to the scale factor used in this benchmark.
10.2.4.5
Verify referential integrity in the database after the initial load. Referential Integrity is a data property that can be
verified by checking that every foreign key has a corresponding primary key.
10.2.4.6
Verify that the qualification and test databases were constructed in the same manner so that correct behavior on the
qualification database is indicative of correct behavior on the test database.
10.2.4.7
Note the method used to populate the database (i.e., DBGEN, modified version of DBGEN, or other method). If
DBGEN was used, note the version number, relelase number, modification number and patch level of DBGEN.
Verify that the version and the release numbers matches the benchmark specification.
10.2.4.8
Verify that storage and processing elements that are not included in the priced configuration are physically removed
or made inaccessible during the performance test.
10.2.4.9
Verify that the database load time is measured according to the requirements.
10.2.5
10.2.5.1
Verify that the driver meets the requirements of Clause 6.2 and Clause 7.3.
10.2.5.2
Verify that the execution rules are followed for the power test.
10.2.5.3
Verify that the queries are executed against the test database.
10.2.5.4
Verify that the execution rules are followed for the throughput test.
10.2.5.5
Verify that a single stream is used for refresh functions in the throughput test and that the required number of refresh
function pairs is executed according to the execution rules.
10.2.5.6
Verify that the query sequencing rules are followed.
10.2.5.7
Verify that the measurement interval for the throughput test is measured as required.
10.2.5.8
Verify that the method used to measure the timing intervals is compliant.
10.2.5.9
Verify that the metrics are computed as required. Note whether Clause 6.4.1.4 concerning the ratio between the
longest and the shortest timing intervals had to be applied.
10.2.5.10 Verify that the reported metrics are repeatable.

Page 132
10.2.6
10.2.6.1
Verify that the composition of the SUT is compliant and that its components will be commercially available software or hardware products according to clause 7 of the Pricing Specification Version 1.
10.2.6.2
Note whether an implementation specific layer is used and verify its compliance with Clause 7.2.4.
10.2.6.3
Verify that the driver's implementation is compliant.
10.2.6.4
Verify that any profile-directed optimization performed by the test sponsor conforms to the requirements of Clause
6.2.9.
10.2.7
10.2.7.1
Verify that major portions of the full disclosure report are accurate and comply with the reporting requirements.
This includes:
The executive summary;
The numerical quantity summary;
The diagrams of both measured and priced configurations;
The block diagram illustrating the database load process.
Page 133
Appendix A: ORDERED SETS

Following are the ordered sets that must be used for sequencing query execution as described in Clause 6.3.5. They
are adapted from Moses and Oakford, Tables of Random Permutations, Stanford University Press, 1963. pp. 52-53.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
Power Test
0
14 2 9 20 6 17 18 8 21 13 3 22 16 4 11 15 1 10 19 5 7 12
Throughput Test
1
21 3 18 5 11 7 6 20 17 12 16 15 13 10 2 8 14 19 9 22 1 4
2
6 17 14 16 19 10 9 2 15 8 5 22 12 7 13 18 1 4 20 3 11 21
3
8 5 4 6 17 7 1 18 22 14 9 10 15 11 20 2 21 19 13 16 12 3
4
5 21 14 19 15 17 12 6 4 9 8 16 11 2 10 18 1 13 7 22 3 20
5
21 15 4 6 7 16 19 18 14 22 11 13 3 1 2 5 8 20 12 17 10 9
6
10 3 15 13 6 8 9 7 4 11 22 18 12 1 5 16 2 14 19 20 17 21
7
18 8 20 21 2 4 22 17 1 11 9 19 3 13 5 7 10 16 6 14 15 12
8
19 1 15 17 5 8 9 12 14 7 4 3 20 16 6 22 10 13 2 21 18 11
9
8 13 2 20 17 3 6 21 18 11 19 10 15 4 22 1 7 12 9 14 5 16
10
6 15 18 17 12 1 7 2 22 13 21 10 14 9 3 16 20 19 11 4 8 5
11
15 14 18 17 10 20 16 11 1 8 4 22 5 12 3 9 21 2 13 6 19 7
12
1 7 16 17 18 22 12 6 8 9 11 4 2 5 20 21 13 10 19 3 14 15
13
21 17 7 3 1 10 12 22 9 16 6 11 2 4 5 14 8 20 13 18 15 19
14
2 9 5 4 18 1 20 15 16 17 7 21 13 14 19 8 22 11 10 3 12 6
15
16 9 17 8 14 11 10 12 6 21 7 3 15 5 22 20 1 13 19 2 4 18
16
1 3 6 5 2 16 14 22 17 20 4 9 10 11 15 8 12 19 18 13 7 21
17
3 16 5 11 21 9 2 15 10 18 17 7 8 19 14 13 1 4 22 20 6 12
18
14 4 13 5 21 11 8 6 3 17 2 20 1 19 10 9 12 18 15 7 22 16
19
4 12 22 14 5 15 16 2 8 10 17 9 21 7 3 6 13 18 11 20 19 1
20
16 15 14 13 4 22 18 19 7 1 12 17 5 10 20 3 9 21 11 2 6 8
21
20 14 21 12 15 17 4 19 13 10 11 1 16 5 18 7 8 22 9 6 3 2
22
16 14 13 2 21 10 11 4 1 22 18 12 19 5 7 8 6 3 15 20 9 17
23
18 15 9 14 12 2 8 11 22 21 16 1 6 17 5 10 19 4 20 13 3 7
24
7 3 10 14 13 21 18 6 20 4 9 8 22 15 2 1 5 12 19 17 11 16
25
18 1 13 7 16 10 14 2 19 5 21 11 22 15 8 17 20 3 4 12 6 9
26
13 2 22 5 11 21 20 14 7 10 4 9 19 18 6 3 1 8 15 12 17 16
27
14 17 21 8 2 9 6 4 5 13 22 7 15 3 1 18 16 11 10 12 20 19
28
10 22 1 12 13 18 21 20 2 14 16 7 15 3 4 17 5 19 6 8 9 11
29
10 8 9 18 12 6 1 5 20 11 17 22 16 3 13 2 15 21 14 19 7 4
30
7 17 22 5 3 10 13 18 9 1 14 15 21 19 16 12 8 6 11 20 4 2
31
2 9 21 3 4 7 1 11 16 5 20 19 18 8 17 13 10 12 15 6 14 22
32
15 12 8 4 22 13 16 17 18 3 7 5 6 1 9 11 21 10 14 20 19 2
33
15 16 2 11 17 7 5 14 20 4 21 3 10 9 12 8 13 6 18 19 22 1
34
1 13 11 3 4 21 6 14 15 22 18 9 7 5 10 20 12 16 17 8 19 2
35
14 17 22 20 8 16 5 10 1 13 2 21 12 9 4 18 3 7 6 19 15 11
36
9 17 7 4 5 13 21 18 11 3 22 1 6 16 20 14 15 10 8 2 12 19
37
13 14 5 22 19 11 9 6 18 15 8 10 7 4 17 16 3 1 12 2 21 20
38
20 5 4 14 11 1 6 16 8 22 7 3 2 12 21 19 17 13 10 15 18 9
39
3 7 14 15 6 5 21 20 18 10 4 16 19 1 13 9 8 17 11 12 22 2
40
13 15 17 1 22 11 3 4 7 20 14 21 9 8 2 18 16 6 10 12 5 19
Page 134
Appendix B: APPROVED QUERY VARIANTS

Following are the approved TPC-H query variants as of the publication date of this version of the specification. As
new query variants may be approved on an on-going basis, implementers are encouraged to obtain a copy of the latest list of approved query variants from the TPC office (see cover page for coordinates).
Some query variants include statements that create temporary tables. In these statements, column data types are designated in angle brackets (e.g., <Integer>) and refer to the list of data types specified in Clause 2.3.1.
- This appendix is also available in machine readable format To obtain a copy of the machine-readable appendices, please contact the TPC (see cover page).
Q8
Variant A (approved 11-Feb-1998)
This variant replaces the CASE statement from the Functional Query Definition with equivalent DECODE() syntax.
The justification for this variant was Clause 3.2.4.3 (d), which allows for vendor-specific syntax that, while not
SQL-92, provides a simple and direct mapping to approved SQL-92 syntax.
select
o_year,
sum(decode(nation, [NATION], volume, 0)) / sum(volume) as mkt_share
from
(
select
l_extendedprice * (1 - l_discount) as volume,
n2.n_name as nation
from
part,
supplier,
lineitem,
orders,
customer,
nation n1,
nation n2,
region
where
and s_suppkey = l_suppkey
and o_custkey = c_custkey
and n1.n_regionkey = r_regionkey
and o_orderdate between date '1995-01-01' and date '1996-12-31'
Page 135
and p_type = '[TYPE]

) all_nations
group by
o_year
order by
o_year;
Q12
This variant replaces the CASE statement from the Functional Query Definition with equivalent DECODE() syntax.
The justification for this variant was Clause 3.2.4.3 (d), which allows for vendor-specific syntax that, while not
SQL-92, provides a simple and direct mapping to approved SQL-92 syntax.
select
l_shipmode,
sum(decode(o_orderpriority, '1-URGENT', 1, '2-HIGH', 1, 0)) as
high_line_count,
sum(decode(o_orderpriority, '1-URGENT', 0, '2-HIGH', 0, 1)) as
low_line_count
from
orders,
lineitem
where
o_orderkey = l_orderkey
and l_shipmode in ('[SHIPMODE1]', '[SHIPMODE2]')
and l_shipdate < l_commitdate
and l_receiptdate >= date '[DATE]'
and l_receiptdate < date '[DATE]' + interval '1' year
group by
l_shipmode
order by
l_shipmode;
Q13
Variant A (approved 5 March 1998)
This variant was required by a vendor which did not support two aggregates in a nested table expression.
create view orders_per_cust[STREAM_ID] (custkey, ordercount) as
select
c_custkey,
count(o_orderkey)
Page 136
from
customer left outer join orders on
and o_comment not like '%[WORD1]%[WORD2]%'
group by
c_custkey;
select
ordercount,
count(*) as custdist
from
orders_per_cust[STREAM_ID]
group by
ordercount
order by
custdist desc,
ordercount desc;
drop view orders_per_cust[STREAM_ID];
Q14
Variant A (approved 5 March 1998)
This variant replaces the CASE statement with the equivalent DECODE() syntax.
select
100.00 * sum(decode(substring(p_type from 1 for 5), 'PROMO',
l_extendedprice * (1-l_discount), 0)) /
sum(l_extendedprice * (1-l_discount)) as promo_revenue
from
lineitem,
part
where
and l_shipdate >= date '[DATE]'
and l_shipdate < date '[DATE]' + interval '1' month;
Q15
This variant was approved because it contains new SQL syntax that is relevant to the benchmark. The SQL3 standard, which was moved to an Approved Committee Draft in May 1996, contains the definition of common table
expressions. TPC-H already makes extensive use of nested table expressions. Common table expressions can be
thought of as shared table expressions or "inline views" that last only for the duration of the query.
with revenue (supplier_no, total_revenue) as (
Page 137
select
l_suppkey,
sum(l_extendedprice * (1-l_discount))
from
lineitem
where
and l_shipdate < date '[DATE]' + interval '3' month
group by
l_suppkey
)
select
s_suppkey,
s_name,
s_address,
s_phone,
total_revenue
from
supplier,
revenue
where
s_suppkey = supplier_no
and total_revenue = (
select
max(total_revenue)
from
revenue
)
order by
s_suppkey;
Page 138
Appendix C: QUERY VALIDATION

Following are the input values and output data for validation of executable query text against the qualification database.
- This appendix is available in machine-readable format only To obtain a copy of the machine-readable appendices, please contact the TPC (see Cover page).
Page 139
Appendix D: DATA AND QUERY GENERATION PROGRAMS

The QGEN (seeClause 3.1.4) and DBGEN (see Clause 5.2.1) programs should be used to generate the executable
query text and the data that populate the TPC-H Databases. These programs produce flat files that can be used by
the test sponsor to implement the benchmark.
- This appendix is available in machine readable format only To obtain a copy of the machine readable appendices, please contact the TPC (see Cover page).
Page 140
Appendix E: SAMPLE EXECUTIVE SUMMARY

This appendix includes a sample Executive Summary. See Clause 9.4 for a detailed description of the required format of the Executive Summary. This sample is provided only as an illustration of the requirements set forth in
Clause 9.4 and Clause 7.3 of the specification. In the event of a conflict between this example and the specification,
the specification shall prevail.
Page 141
TPC-H Rev. 2.3.0
My Logo
My Server
Total System Cost
Report Date:
November 11, 2005
Composite Query per Hour Metric
18725.9
$523,764 USD
QphH@300GB
Database Size
Database Manager
Operating System
300GB
My DB
My OS
Price / Performance
$27.97 USD
$ / QphH@300GB
Other
Availability Date
Software
November 11, 2005
Q1
Q2
Q3
Power Test
Q4
327.3
50.6
Throughput Test
Q5
Q6
..Geomet ric M ean of Power t est

__Arit hmet ic M ean of Throughput Test
Q7
Q8
Q9
Q10
Q11
Q12
Q13
Q14
Q15
Q16
Q17
Q18
Q19
Q20
Q21
Q22
UF2
UF1
0.0
200.0
400.0
Database Load Time = 1:42:55

RAID (Base tables only): Y
600.0
800.0
1000.0
1200.0
1400.0
1600.0
Load Included Backup: Y

Total Data Storage / Database Size = 23.51
RAID (Base tables and auxiliary data structures): Y
RAID (All): Y
System Components System Total
Nodes:
Processors (Dual-Core My Processor Model ABC(3.0GHz/1MB L2 cache, 8MB L3 cache):
Cores:
Threads:
Memory:
OS Disk Drives (36GB 15krpm HDD Ultra320):
Network Interfaces (on-board GigE):
Host Bus Adapters (on-board):
Storage Switch (My SAN Switch):
Storage Subsystem (My Storage Array):
Storage Subsystem Disk Drives (36GB 15krpm HDD Ultra320):
Total Storage :
8
8
16
16
64GB
16
16
16
2
16
208
7062.11GB
Per Node
n.a.
1
2
2
8GB
2
2
2
n.a.
n.a.
n.a.
n.a.
Page 142
TPC-H Rev. 2.3.0
My Log
My Server
Description
Part Number Brand Pricing
Report Date:
Unit Price
11-Nov-05
Extended
Price
Qty
3 yr. Maint.
Price
Server Hardware
My Co. Server
aaaaaa-b1
3,152
25,216
My Co. 4GB Reg PC3200 2x2GB Memory
aaaaaa-b2
2,049
16
32,784
36GB 15Krpm U320 UNI HDD
aaaaaa-b3
299
16
4,784
My Co. Fiber Channel Adapter
aaaaaa-b4
599
4,792
My Co. Care Pack 3-year, 4-hour, 24x7
aaaaaa-b5
295
My Co. unassembled Rack
aaaaaa-b6
689
1,378
My Co. System Server Enclosure
aaaaaa-b7
1,499
1,499
My Co. Interconnect Kit
aaaaaa-b8
4,399
4,399
My Co. Pwr Spply Kit
aaaaaa-b9
2,299
2,299
My Co Enclosure Care Pack 3yr 4hr 24x7
aaaaaa-b10
737
My Co. UPS
aaaaaa-b11
749
My Co. CAT5 KVM USB 1 Pack Interface Adapter
aaaaaa-b12
99
99
My Co. CRT Monitor
aaaaaa-b13
149
149
USB Optical Mouse cbt/slvr USB 2-Button Optical Scroll Mouse
aaaaaa-b14
12
12
Standard Keyboard USB
Subtotal
2,360
737
1,498
78,914
3,097
Storage
My Co. Storage Array
aaaaaa-b18
6,995
16
111,920
36GB, 15krpm HDD Ultra320
aaaaaa-b19
299
192
57,408
2Gb SFF-SW Trncvr Kit

5m LC to LC Cable Kit
aaaaaa-b20
aaaaaa-b21
1
1
199
82
32
32
6,368
2,624
My Co. Storage Array Support 3YR 24x7 4HR
aaaaaa-b22
3,538
16
My Co. SAN Switch

My Co. SAN Swithc 3Y 4H 24x7 HW support
aaaaaa-b23
aaaaaa-b24
aaaaaa-b25
1
1
9,990
4,141
2
2
19,980
Subtotal
198,300
56,608
8,282
64,890
Software
My DB
aaaaaa-b27
My DB
10,000
12*
120,000
My DB Cluster software for 3 Years
aaaaaa-b28
My DB
5,000
12*
60,000
My DB Storage Softwar 3 Years
aaaaaa-b29
My DB
2,500
12*
30,000
My DB Support Package for 3 Years
aaaaaa-b30
My DB
16,000
My Co. OS
aaaaaa-b31
2,157
17,256
Subtotal
227,256
48,000
My Co. Large Purchase and Net30 discount
(47,115)
My DB Discount
(38,700)
Total
Pricing: 1- My Co Direct: 832-752-7583 at 16% discount.

Pricing: 2- My DB at 15% discount
Included
48,000
(10,878)
418,655
105,109
3-Year Cost of Ownership:
$523,764
QphH Rating:
18725.9
$ / QphH@300GB:
$27.97
Prices used in TPC benchmarks reflect the actual prices a customer would pay for a one-time purchase of the stated components. Individually negotiated discounts are not permitted. Special prices based on assumptions about past or future.purchases are not
permitted. All discounts reflect standard pricing policies for the listed components. For complete details, see the pricing sections of the TPC benchmark specifications. If you find that the stated prices are not available according to these terms, please inform the
TPC at [email protected]. Results independently audited by My Auditor Inc. Thank you
Page 143
TPC-H Rev. 2.3.0
My Logo
My Server
Report Date:
November 11, 2005
Numerical Quantities
Measurement Results
Database Scale Factor
Total Data Storage / Database Size
Start of Database Load
End of Database Load
Database Load Time
= 300
= 23.51
= 10/27/2005 16:10:35
= 10/27/2005 17:31:19
= 01:20:44
Query Streams for Throughput Test

TPC-H Power
TPC-H Throughput
TPC-H Composite Query-per-Hour Metric (QphH@300GB)
Total System Price Over 3 Years
TPC-H Price/ Performance Metric ($/QphH@300GB)
=6
= 21338.2
= 16433.4
= 18725.9
= $523,764
= $27.97
Measurement Intervals
Measurement Interval in Throughput Test
= 8675 seconds
Duration of Stream Execution

Seed
Power Stream
1027173119
Throghput Stream
Seed
1027173120
1027173121
1027173122
1027173123
1027173124
1027173125
RF1 Start Time

RF1 End Time
10/27/2005 20:53:10
10/27/2005 20:54:00
Query Start Time
Query End Time
10/27/2005 21:25:20
10/27/2005 23:29:29
10/27/2005 21:25:20
10/27/2005 23:40:01
10/27/2005 21:25:20
10/27/2005 23:31:38
10/27/2005 21:25:20
10/27/2005 23:04:49
10/27/2005 21:25:20
10/27/2005 23:19:16
10/27/2005 21:25:20
10/27/2005 23:26:51
Query Start Time

RF2 Start Time
Query End Time
RF2 End Time
10/27/2005 20:54:00 10/27/2005 21:24:34
10/27/2005 21:24:34 10/27/2005 21:25:19
Duration
2:16:08
2:17:35
2:19:16
2:20:45
2:22:39
2:24:35
RF1 Start Time

RF1 End Time
10/27/2005 23:40:01
10/27/2005 23:40:49
10/27/2005 23:41:28
10/27/2005 23:42:15
10/27/2005 23:42:55
10/27/2005 23:43:48
10/27/2005 23:44:36
10/27/2005 23:45:19
10/27/2005 23:46:05
10/27/2005 23:46:49
10/27/2005 23:47:59
10/27/2005 23:49:01
Duration
0:32:09
RF2 Start Time

RF2 End Time
10/27/2005 23:40:49
10/27/2005 23:41:28
10/27/2005 23:42:15
10/27/2005 23:42:55
10/27/2005 23:43:48
10/27/2005 23:44:36
10/27/2005 23:45:19
10/27/2005 23:46:05
10/27/2005 23:46:49
10/27/2005 23:47:59
10/27/2005 23:49:01
10/27/2005 23:49:55
Page 144
TPC-H Rev. 2.3.0
My Logo
My Server
Report Date:
November 11, 2005
TPC-H Timing Intervals (in seconds)

Query
Q1
Q2
Q3
Q4
Q5
Q6
Q7
Q8
Q9
Q10
Q11
Q12
Q13
Q14
Q15
Q16
Q17
Q18
Q19
Q20
Q21
Q22
UF1
UF2
Stream 0
202.8
35.4
19.9
12.9
56.9
13.6
48.6
65.4
221.3
23.2
61.9
85.6
57.4
4.5
22.4
40.9
84.1
260.0
101.4
43.0
335.4
37.7
49.9
45.5
Stream 1
404.6
65.2
74.3
26.5
221.1
32.7
203.7
208.9
920.2
474.9
115.1
238.3
201.3
43.0
154.0
345.0
212.7
1807.9
332.4
91.6
1178.2
97.8
48.6
39.0
Stream 2
234.9
113.9
18.2
10.9
452.4
32.4
327.8
161.8
2257.4
830.0
53.8
203.2
172.9
64.8
200.0
907.7
439.9
806.1
274.0
45.2
281.8
191.5
46.7
40.3
Stream 3
506.7
96.6
52.3
58.9
321.3
45.7
273.8
203.2
992.9
78.1
189.6
135.6
188.9
27.3
215.4
260.8
322.0
1501.7
203.9
156.3
1678.4
29.2
52.8
48.1
Stream 4
427.7
47.2
66.2
28.7
337.5
18.1
355.8
149.4
437.5
106.5
116.0
178.6
167.9
37.3
133.6
162.5
145.7
1114.1
143.0
224.6
1445.8
124.8
42.9
46.2
Stream 5
500.2
87.8
95.8
65.0
373.5
32.2
190.8
155.1
799.4
43.8
169.3
302.2
209.7
58.0
154.3
320.2
339.6
1436.3
151.4
63.3
1186.8
101.5
43.6
70.2
Stream 6
433.1
156.4
104.1
28.8
233.8
61.0
128.8
257.1
790.1
79.9
157.7
227.2
187.6
60.8
247.0
271.7
315.2
1371.5
264.2
431.2
1398.4
85.9
61.7
54.7
Min Qi
234.9
47.2
18.2
10.9
221.1
18.1
128.8
149.4
437.5
43.8
53.8
135.6
167.9
27.3
133.6
162.5
145.7
806.1
143.0
45.2
281.8
29.2
42.9
39.0
Max Qi
506.7
156.4
104.1
65.0
452.4
61.0
355.8
257.1
2257.4
830.0
189.6
302.2
209.7
64.8
247.0
907.7
439.9
1807.9
332.4
431.2
1678.4
191.5
61.7
70.2
Avg Qi
417.9
94.5
68.5
36.4
323.3
37.0
246.8
189.3
1032.9
268.9
133.6
214.2
188.0
48.5
184.1
378.0
295.8
1339.6
228.1
168.7
1194.9
105.1
49.4
49.7
Page 145
Page 146
Page 147
Page 148
Appendix F: REFERENCE DATA SET
This appendix contains sample dbgen and qgen data (reference data set) and the command lines/scripts
used to generate this data by the TPC.
Base Data Set
The base data set contains sample data for all tables at all scale factors. For each scale factor 5 files of
tables lineitem, orders, part, partsupp, customer and supplier are included. For tables nation and region
all data is included due to their limited size.
Insert Data Set
The insert data set contains sample data for tables lineitem and orders at all scale factors. For all scale
factors and each of the update sets 1, 75 and 150 100 files for lineitem and 100 files for orders are
included.
Delete Data Set
The delete data set contains sample data for tables lineitem and orders at all scale factors. For each
scale factor 100, 300, 1000, 3000, 10000, 30000, 100000 and each of the update sets 1, 75 and 150 100
files are included. For scale factor 1 and each of the update sets 1, 75 and 150 94 files are included.
Qgen Data Set
The qgen data set contains 150 files with query substitutions values for all 22 queries for each scale factor as generated with qgen. Each file uses a different seed.
- This appendix is available in machine readable format only To obtain a copy of the machine readable appendices, please contact the TPC (see Cover page).
Page 149

tpch2 8 0

Uploaded by

Copyright:

Available Formats

tpch2 8 0

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

tpch2 8 0

Uploaded by

Copyright:

Available Formats

TPC BENCHMARKTM H

Transaction Processing Performance Council (TPC)

TPC Benchmark H Standard Specification Revision 2.8.0

Mail ballot draft for Standard Specification

First minor revision of the Specification

Clarification about Primary Keys

Additions for EOL of hardware in 8.6

Mail ballot draft 3 year maintenance pricing

Adding scale factors 30TB and 100TB

Adding Pricing Specification 1.0.0

Changing pricing precision to cents and processor definition

TPC Benchmark H Standard Specification Revision 2.8.0

dbgen bug fixes in parallel data generation,

Added Clause 7.2.3.1 about software license

Editorial correction in Clause 2.1.3.3

Change substr into substring in Clause 2.25.2,

Incorporate BUG fix 595 of qgen

Add wording to allow substitutions in Clause

TPC Benchmark H Standard Specification Revision 2.8.0

Revisions to the Full Disclosure Report . . . . . . . . . . . . . . . . . . . . . . 128

. . . . . . . . . . . . . . . SAMPLE EXECUTIVE SUMMARY141

Appendix F REFERENCE DATA SET . . . . . . . . . . . . . . . . . . . . . . . . . 149

TPC Benchmark H Standard Specification Revision 2.8.0

Examine large volumes of data;

Execute queries with a high degree of complexity;

Give answers to critical business questions.

Give answers to real-world business questions;

Are far more complex than most OLTP transactions;

Include a rich breadth of operators and selectivity constraints;

The TPC-H operations are modeled as follows:

General Implementation Guidelines

Are generally available to users;

Is the implementation generally available, documented, and supported?

General Measurement Guidelines

The approach is an accepted engineering practice or standard;

The approach does not enhance the result;

TPC Benchmark H Standard Specification Revision 2.8.0

TPC Benchmark H Standard Specification Revision 2.8.0

2: LOGICAL DATABASE DESIGN

Business and Application Environment

They have a high degree of complexity;

They use a variety of access

They are of an ad hoc nature;

patterns; theyexamine a large percentage of the available data;

They all differ from each other;

They contain query parameters that change across query executions.

Pricing and promotions;

Supply and demand management;

Profit and revenue management;

Customer satisfaction study;

Market share study;

TPC Benchmark H Standard Specification Revision 2.8.0

Database Entities, Relationships, and Characteristics

TPC Benchmark H Standard Specification Revision 2.8.0

SF*200,000 are populated

variable text, size 55

fixed text, size 10

variable text, size 25

fixed text, size 10