Scribd Logo
Upload

Welcome to Scribd!

  • 31 views

    Locking & Blocking Made Simple: Our Agenda

    Uploaded by

    Rama Udaya
    This document discusses locking and blocking in multi-user databases. It begins with background on Joe Webb and an overview of locking, blocking, and transaction isolation levels. The bulk of the document covers locking types and lock compatibility, techniques for controlling locks like hints and isolation levels, and tools for monitoring locking and blocking. It aims to explain the nature of locking and blocking in databases and provide strategies for avoiding excessive locking.

    Copyright:

    Attribution Non-Commercial (BY-NC)
    Download as PDF, TXT or read online from Scribd

    Locking & Blocking Made Simple: Our Agenda

    Uploaded by

    Rama Udaya
    31 views10 pages
    This document discusses locking and blocking in multi-user databases. It begins with background on Joe Webb and an overview of locking, blocking, and transaction isolation levels. The bulk of the document covers locking types and lock compatibility, techniques for controlling locks like hints and isolation levels, and tools for monitoring locking and blocking. It aims to explain the nature of locking and blocking in databases and provide strategies for avoiding excessive locking.

    Original Description:

    Original Title

    Link Click

    Copyright

    © Attribution Non-Commercial (BY-NC)

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    This document discusses locking and blocking in multi-user databases. It begins with background on Joe Webb and an overview of locking, blocking, and transaction isolation levels. The bulk of the document covers locking types and lock compatibility, techniques for controlling locks like hints and isolation levels, and tools for monitoring locking and blocking. It aims to explain the nature of locking and blocking in databases and provide strategies for avoiding excessive locking.

    Copyright:

    Attribution Non-Commercial (BY-NC)
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    31 views10 pages

    Locking & Blocking Made Simple: Our Agenda

    Uploaded by

    Rama Udaya
    This document discusses locking and blocking in multi-user databases. It begins with background on Joe Webb and an overview of locking, blocking, and transaction isolation levels. The bulk of the document covers locking types and lock compatibility, techniques for controlling locks like hints and isolation levels, and tools for monitoring locking and blocking. It aims to explain the nature of locking and blocking in databases and provide strategies for avoiding excessive locking.

    Copyright:

    Attribution Non-Commercial (BY-NC)
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    of 10

    9/11/2009

    Locking & Blocking Made Simple

    Joe Webb Microsoft SQL Server MVP WebbTech Solutions, LLC [email protected]

    Our Agenda
    The Nature of Multi-User Databases The Basics of Locking and Blocking Techniques for Controlling Locking

    A Bit of Background
    Joe Webb Started in IT in 1993 after receiving a BSEE & MBA from Auburn University Founded WebbTech Solutions in 1996 Providing consulting, mentoring, & technical training Began working with SQL Server in 1996 with version 6.0 5-time recipient of the Microsoft MVP Award for SQL Server Author or Coauthor of 3 books Served 6 years on the Board of Directors for PASS (sqlpass.org) Served 2 years on the Executive Committee as VP of Marketing and EVP of Finance Delivered more than 50 sessions at over 20 conferences in North America & Europe

    9/11/2009

    The Nature of Multi-User Databases

    An Example
    Total = $6,200

    Pay $800

    Balance = $200

    Customer_ID 123 124 125 126

    Balance $200.00 $1,000.00 $1,500.00 $2,000.00 $2,500.00

    The ACID Test

    Atomicity: Transactions are all or nothing

    Consistency: Only valid data is saved

    Isolation: Transactions do not affect each other

    Durability: Written data will not be lost

    9/11/2009

    Locking & Blocking

    Ensures that resources are not being simultaneously used by different processes in conflicting ways Provides consistency Prevents conflict

    Results when one process must wait for another process to release a resource Decreases transactional throughput, increases wait times, etc

    Locking

    Types of Locks
    Lock Type Shared (S) Update (U) Description Indicates that the resource is being read by a process Indicates that a process is going to update the resource. Only one Update lock can exist at a time for a resource. Before writing, its converted to an Exclusive lock Used to prevent multiple processes from attempting the same resource at one time. Used to establish lock hierarchy. They are placed before lower level locks for efficiency and to protect the lower level locks Protects the definition of a resource during data manipulation or prevents data manipulation when a DDL command is executing Used when copying large amounts of data into a table Protects a range of key values for SERIALIZABLE operations

    Exclusive (X) Intent (I) Schema (Sch)

    Bulk Update Key Range

    Lockable Resources
    Resource
    Database File Object Page Key Extent RID Application Metadata HOBT Allocation unit

    Blocking

    9/11/2009

    Lock Compatibility
    Shared Shared Update Exclusive Update Exclusive

    = Ok = Wait

    Lock Granularity

    Table Partition Page Row

    Lock Escalation
    SQL Server attempts to be as granular as possible to increase concurrency Lock Escalation is the process of converting many lowerlevel locks to fewer higher level locks Escalations occur when:
    The number of locks for a table scan is greater than approximately 5,000 The amount of non-AWE memory used for locks exceeds 24% and the locks setting is zero The amount of non-AWE memory used for locks exceed 40% and the locks setting is non-zero

    Escalation decreases the amount of server resources required to keep track of locks while increasing the possibility of concurrency conflicts

    9/11/2009

    Techniques for Controlling Locks


    Using Granularity Hints Using Transaction Isolation Levels

    Granularity Hints
    Hint NOLOCK PAGLOCK ROWLOCK TABLOCK TABLOCKX UPDLOCK XLOCK Description Instructs SQL Server to disregard locks placed on resources, allowing dirty reads Places locks at the page level, rather than row or table level Places locks on rows rather than pages or tables Places locks on a table rather than at a lower level Places an exclusive lock on a table Places an update lock on a resource Indicates that an exclusive lock is to be placed on the specified resource; maybe used with PAGLOCK, ROWLOCK, and TABLOCK

    Transaction Isolation Level

    Read Uncommitted Read Committed Repeatable Read Serializable Snapshot

    Transaction Isolation Levels

    9/11/2009

    Transaction Isolation Levels


    Read Uncommitted
    Instructs SQL Server to completely ignore normal locking activities Does not place read locks on resources when reading Does not observe write locks on resources when reading Allows dirty reads reading uncommitted data from other transactions Most beneficial for lookup tables that seldom change or for Data Warehouse applications Can also be invoked using the WITH (NOLOCK) hint

    Transaction Isolation Levels


    Read Committed
    Reads only data that has been committed to the database If resources are exclusively locked, the process will wait indefinitely for the lock to be released Prevents dirty reads but allows non-repeatable reads and phantom reads Places a shared lock on resources during reads Default transaction isolation level for SQL Server Can control how long a transaction will wait for a resource using the LOCK_TIMEOUT option for the connection When using the LOCK_TIMEOUT option, client applications must catch the error raised when a query times out

    Transaction Isolation Levels


    Repeatable Read
    Designed to ensure that read data is not changed by other processes during a transaction Places a shared lock on all read resource; the lock persists until the transaction is completed or rolled back Your transaction can see its own changes to data Increases the duration of read locks and hence has and adverse affect on concurrency Allows phantom reads

    9/11/2009

    Transaction Isolation Levels


    Serializable
    Places a shared lock on all resources read during a transaction; the lock persists until the transaction is committed or rolled back Other transactions cannot insert rows into the table that would change the result of this transaction Implemented with Key Range locks Prevents phantom reads but has a significantly adverse affect on concurrency Can also be invoked using the WITH (HOLDLOCK) hint

    Transaction Isolation Levels


    Read Committed Snapshot
    Implemented by setting the READ_COMMITTED_SNAPSHOT ON database setting, causes READ COMMITTED isolation level to use row versioning Provides statement level snapshots using row versioning; versioning information is stored in data pages within tempdb Read operations retrieve the latest copy of data from the version store when the statement is started Write operations perform like READ COMMITTED using update and exclusive locks

    Transaction Isolation Levels


    Snapshot Isolation
    Implemented with the ALLOW_SNAPSHOT ISOLATION ON database option is set, allows transaction isolation level to be set to SNAPSHOT Provides transaction level isolation using row versioning Row version information is kept in tempdb and data is read as of the start of the transaction using the current transaction id Write operations use an optimistic locking (ie no update locks are placed) which can cause update conflicts to occur The likelihood of conflicts can be reduced by using the WITH (UPDLOCK) hint

    9/11/2009

    Avoiding Excessive Locking


    Database Design
    Normalize the database Consider column partitioning in cases where diverse access methods are used Consider horizontal partitioning to decrease blocking exposure and provide multiple objects to be accessed

    Indexing
    Use narrow keys for indexes Create adequate indexes for reading data to reduce the number of rows to be read Consider including additional columns in nonclustered indexes to convert them to covering indexes

    Avoiding Excessive Locking


    Optimization
    Optimize queries, resolving the long running queries that consume resources and keep locks open Update statistics Review query plans to ensure the optimal plans are being used Keep transactions as short as possible Increase non-AWE memory using the 3GB switch Reduce the occurrences of Key Lookups Avoid using resource intensive programming techniques such as cursors, etc.

    Avoiding Excessive Locking


    Transaction Isolation Levels
    Select the least restrictive possible for the given set of tasks For example, choose READ UNCOMMITTED for Business Intelligence scenarios Consider implementing row versioning isolation Readers do not block writer; writers do not block readers Overall number of locks is reduced thus reducing system overhead

    9/11/2009

    Tools to Monitor Locking & Blocking


    DMVs & System Views sys.dm_tran_locks - returns lock manager resources and current requests for these resources for the SQL Server instance sys.partitions contains information about system resources that may be locked sys.dm_exec_sessions returns information about the active sessions Windows Performance Monitor The SQL Server: Locks object provides information such as Average Wait time (ms), Lock Requests / sec, and Lock Waits / sec

    Tools to Monitor Locking & Blocking


    SQL Profiler Use the Blocked Process Report event class Must define how long a process is blocked to be considered blocked using the 'blocked process threshold configuration option sp_blocker_pss08 Procedure Stored procedure available from KB 271509 that queries underlying system tables and views to report blocking. Available for SQL Server 2000, 2005, and 2008 Data Collection in SQL Server 2008 The data collector collects information from the DMVs and preserves that information in a database that can be queried to look for trends over time

    Additional Resources
    http://www.sql-server-performance.com http://msdn.microsoft.com/enus/library/ms191168.aspx http://www.simpletalk.com/content/article.aspx?article=671 http://support.microsoft.com/kb/271509 http://support.microsoft.com/kb/271509 Inside SQL Server books http://www.webbtechsolutions.com/blog http://www.twitter.com/joewebb

    9/11/2009

    Questions?

    Thank you!
    [email protected] http://www.webbtechsolutions.com/blog http://www.twitter.com/joewebb

    10

    You might also like