Distributed Systems - Concepts and Design

Distributed Garbage Collection
An Overview
Presented by
Dotan Adler
Copyright, 1996 © Dale Carnegie & Associates, Inc.

Representation Outline
• Present the need for Distributed GC

• Present the Distributed Model
• Present the problem of Distributed GC
• Present some outlines of solutions to the
problem (one Direct, and two Indirect)
• Conclusion
‫תשע"ט‬/‫ניסן‬/‫כ"ד‬ Distributed Garbage Collection 2

The need for a Distributed GC
• Where do we need Distributed GC ?

– Distributed applications (DCOM, CORBA, etc ..)
– Internet applications (Java extensions)
– Distributed file servers
– HTML pages & links

The Distributed Computing Model
A Distributed System is a set of Autonomous
systems connected by a network.
IBM Compatible
Cray Supercomputer
Communication
Network
Mac Classic
IBM PS/2
Laptop computer

The Classical GC Problem

• Each computer in a distributed system has :

– A local store (memory or disk)
– A stack
– Local running programs (or RPCs)
– A GC algorithm (a part of a global GC algorithm)
• Each computer has access only to local store
• Access to remote store is achieved by message
passing

• Because there is no global address space,

references to remote cells are necessarily
indirect
A Entry Communication Exit B

Recotd Network Recotd
a b

Problems With Distributed GC
• Unreliability (Relaxed)
– Duplications of messages
– Out of order delivery
– Communication fail
• Latency - the elapsed time between the issue of a
task and when it is executed is undeterministic
• Synchronization - it takes a lot of resources to
synchronize all computers in a network

Direct Approach
• Direct approach means to identify the garbage

and mark it, so that it could be removed
• Reference counting is a direct GC approach
since it marks garbage with count equals to 0

Reference Counting
• Every object has a reference count

• Inc(@a) message increases a’s count
• Dec(@a) decreases the count
• “Inc”, and “Dec” messages
are sent to the CPU which
hosts the object inc(@b)
• Deletion occurs when cp(@b)
reference equals 0
A B C
Reference Counting (Cont’d)
• Problem : If A or C were responsible for

incrementing b’s count, then latency is an
important factor
A B
@b inc(@b)
b del(@b)
inc(@b)
del(@b)
C
cp(@b) cp(@b)
@b
A B C A B C
A is responsible for inc C is responsible for inc

• Solution : When A duplicates @b, it first sends

an ack-request to B and a copy message to C
• B will not accept any
request after it gets an ack(@b)
ack_req until it sends
an ack to C cp(@b)
• C will not use @b until
it get ack(@b) ack_req(@b,C)
A B C

• Problems
– Does not destroy inter-site cyclic references
– Relies on blocking of operations

Indirect Approach
• Indirect approach means to identify all the live

objects first, and then reclaim all space not used
by “live” objects
• Examples of indirect GC algorithms :
– Mark-scan
– Generation scavenging
– Dijkstra’s 3-color algorithm

Centralized Indirect Solution
• Let every site run a GC locally on it’s private

memory.
RCT = remote cell table ERT = exit reference table

Centralized Indirect Solution (Cont)
• Still we have problems with inter-site cycles.

• For this we add a centralized service :
– Whenever a local site finishes it’s GC it sends a list
of RCT to ERT paths, which are inaccessible from
the roots to the centralized service.
– The centralized service can then find dead inter-site
cycles, and report them to the sites.

Cell Migration
• We can solve the inter-site cycles problem by
migrating cells, so that inter-site cycles become
local cycles.
• We need to define an order, among the sites, so
that we won’t get a cyclic migration. This way a
cell migrates only to an inferior site.
A B A B A B
a c a c a b
b
b
C C C
c

Cell Migration
• Problems:
– Copy operations could be very slow - especially
when working with big networks, and big data
structures
– An order between sites must be set prior to the
running of the algorithm

Empirical Results
• Little work has been done in the field of DGC

because it is hard to compare the execution of
two distributed systems. (delay times,
synchronization, etc …)
• Tests, however, show that Direct methods give
better pause times than Indirect methods.

Future Directions of research
• Use some hybrid methods (Direct & Indirect)

• Try to create a framework for testing
Distributed GC algorithms

Conclusions
• Distributed GC adds a new dimesion to old-

style GC. It adds the dimension of asynchronity
& latency.
• Alltought standard GCs suffer from the same
problems (memory leaks, pause time), creating
a robust DGC is much more complicated.

The End

Distributed Systems - Concepts and Design

Uploaded by

Copyright:

Available Formats

Distributed Systems - Concepts and Design

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

Distributed Systems - Concepts and Design

Uploaded by

Copyright:

Available Formats

Distributed Garbage Collection

Copyright, 1996 © Dale Carnegie & Associates, Inc.

• Present the need for Distributed GC

‫תשע"ט‬/‫ניסן‬/‫כ"ד‬ Distributed Garbage Collection 2

• Where do we need Distributed GC ?

‫תשע"ט‬/‫ניסן‬/‫כ"ד‬ Distributed Garbage Collection 3

‫תשע"ט‬/‫ניסן‬/‫כ"ד‬ Distributed Garbage Collection 4

‫תשע"ט‬/‫ניסן‬/‫כ"ד‬ Distributed Garbage Collection 5

• Each computer in a distributed system has :

‫תשע"ט‬/‫ניסן‬/‫כ"ד‬ Distributed Garbage Collection 7

• Because there is no global address space,

A Entry Communication Exit B

‫תשע"ט‬/‫ניסן‬/‫כ"ד‬ Distributed Garbage Collection 8

‫תשע"ט‬/‫ניסן‬/‫כ"ד‬ Distributed Garbage Collection 9

• Direct approach means to identify the garbage

‫תשע"ט‬/‫ניסן‬/‫כ"ד‬ Distributed Garbage Collection 10

• Every object has a reference count

• Deletion occurs when cp(@b)

• Problem : If A or C were responsible for

A is responsible for inc C is responsible for inc

‫תשע"ט‬/‫ניסן‬/‫כ"ד‬ Distributed Garbage Collection 12

• Solution : When A duplicates @b, it first sends

‫תשע"ט‬/‫ניסן‬/‫כ"ד‬ Distributed Garbage Collection 13

‫תשע"ט‬/‫ניסן‬/‫כ"ד‬ Distributed Garbage Collection 14

• Indirect approach means to identify all the live

‫תשע"ט‬/‫ניסן‬/‫כ"ד‬ Distributed Garbage Collection 15

• Let every site run a GC locally on it’s private

RCT = remote cell table ERT = exit reference table

‫תשע"ט‬/‫ניסן‬/‫כ"ד‬ Distributed Garbage Collection 16

• Still we have problems with inter-site cycles.

‫תשע"ט‬/‫ניסן‬/‫כ"ד‬ Distributed Garbage Collection 17

‫תשע"ט‬/‫ניסן‬/‫כ"ד‬ Distributed Garbage Collection 18

‫תשע"ט‬/‫ניסן‬/‫כ"ד‬ Distributed Garbage Collection 19

• Little work has been done in the field of DGC

‫תשע"ט‬/‫ניסן‬/‫כ"ד‬ Distributed Garbage Collection 20

• Use some hybrid methods (Direct & Indirect)

‫תשע"ט‬/‫ניסן‬/‫כ"ד‬ Distributed Garbage Collection 21

• Distributed GC adds a new dimesion to old-

‫תשע"ט‬/‫ניסן‬/‫כ"ד‬ Distributed Garbage Collection 22

‫תשע"ט‬/‫ניסן‬/‫כ"ד‬ Distributed Garbage Collection 23

You might also like