[
https://issues.apache.org/jira/browse/DERBY-2911?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Knut Anders Hatlen updated DERBY-2911:
--------------------------------------
Attachment: d2911-6.stat
d2911-6.diff
Attaching a new patch (d2911-6.diff) which implements a replacement algorithm
using the interface and synchronization model described in the previous
comments. The main part of the patch is the new ClockPolicy class and its
rotateClock() method. I have done my best to state all requirements about
synchronization and lock order in comments, so I hope it is possible for others
to understand the code...
I manually edited modules.properties to enable the new buffer manager and ran
the full regression test suite. suites.All ran successfully, whereas derbyall
had one failure. The failure in derbyall was unit/T_RawStoreFactory.unit which
has been mentioned before. This is an expected failure until reuse of invalid
entries has been enabled for caches whose size is smaller than the maximum size.
I have not run any performance tests on this last patch, but I will do so. The
performance test attached to this issue doesn't test the replacement algorithm
since it creates just a small database. I will therefore see if I can run some
other tests with different buffer sizes and also test it with update load, for
instance using the test client attached to DERBY-1961.
What's left to do:
- reuse Cacheable from invalid entries instead of growing the clock when size
< maxSize (will fix the failure in unit/T_RawStoreFactory.unit)
- implement a background cleaner
- shrink the cache if it exceeds the maximum size
> Implement a buffer manager using java.util.concurrent classes
> -------------------------------------------------------------
>
> Key: DERBY-2911
> URL: https://issues.apache.org/jira/browse/DERBY-2911
> Project: Derby
> Issue Type: Improvement
> Components: Performance, Services
> Affects Versions: 10.4.0.0
> Reporter: Knut Anders Hatlen
> Assignee: Knut Anders Hatlen
> Priority: Minor
> Attachments: d2911-1.diff, d2911-1.stat, d2911-2.diff, d2911-3.diff,
> d2911-4.diff, d2911-5.diff, d2911-6.diff, d2911-6.stat,
> d2911-entry-javadoc.diff, d2911-unused.diff, d2911-unused.stat, d2911perf.java
>
>
> There are indications that the buffer manager is a bottleneck for some types
> of multi-user load. For instance, Anders Morken wrote this in a comment on
> DERBY-1704: "With a separate table and index for each thread (to remove latch
> contention and lock waits from the equation) we (...) found that
> org.apache.derby.impl.services.cache.Clock.find()/release() caused about 5
> times more contention than the synchronization in LockSet.lockObject() and
> LockSet.unlock(). That might be an indicator of where to apply the next push".
> It would be interesting to see the scalability and performance of a buffer
> manager which exploits the concurrency utilities added in Java SE 5.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
