[
https://issues.apache.org/jira/browse/SOLR-17040?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17776842#comment-17776842
]
Alex Deparvu edited comment on SOLR-17040 at 10/18/23 6:29 PM:
---------------------------------------------------------------
I am making heavy use of the CloudIndexing benchmark and have some interesting
(if not confusing) results to share.
In short: *without profiler enabled the change comes out a bit under main
branch BUT with profiler enabled there is a 15% boost in throughput*. I am
still trying to wrap my head around why this is happening and the results are
not reproducible without profiling enabled
* Regular benchmark run (main looks a bit better 32k vs 31k):
{noformat}
main
Benchmark (nodeCount) (numReplicas) (numShards) (preGenerate)
(useSmallDocs) Mode Cnt Score Error Units
CloudIndexing.indexDoc 1 1 1 50000
true thrpt 4 32030.118 ± 449.769 ops/s
patch
Benchmark (nodeCount) (numReplicas) (numShards) (preGenerate)
(useSmallDocs) Mode Cnt Score Error Units
CloudIndexing.indexDoc 1 1 1 50000
true thrpt 4 31215.091 ± 671.106 ops/s
{noformat}
* Async profiler enabled (numbers basically reverse, patch looks a lot better
26k vs 23k)
{noformat}
main
Benchmark (nodeCount) (numReplicas) (numShards)
(preGenerate) (useSmallDocs) Mode Cnt Score Error Units
CloudIndexing.indexDoc 1 1 1
50000 true thrpt 4 22988.945 ± 1699.597 ops/s
patch
Benchmark (nodeCount) (numReplicas) (numShards)
(preGenerate) (useSmallDocs) Mode Cnt Score Error Units
CloudIndexing.indexDoc 1 1 1
50000 true thrpt 4 26276.781 ± 1308.604 ops/s
{noformat}
benchmark command for reference:
{noformat}
./jmh.sh -f 1 -r 300s -to 450s CloudIndexing -p nodeCount=1 -p numShards=1 -p
numReplicas=1 -jvmArgsAppend -Dsolr.http1=true
{noformat}
was (Author: alex.parvulescu):
I am making heavy use of the CloudIndexing benchmark and have some interesting
(if not confusing) results to share:
* Regular benchmark run (main looks a bit better 32k vs 31k):
{noformat}
main
Benchmark (nodeCount) (numReplicas) (numShards) (preGenerate)
(useSmallDocs) Mode Cnt Score Error Units
CloudIndexing.indexDoc 1 1 1 50000
true thrpt 4 32030.118 ± 449.769 ops/s
patch
Benchmark (nodeCount) (numReplicas) (numShards) (preGenerate)
(useSmallDocs) Mode Cnt Score Error Units
CloudIndexing.indexDoc 1 1 1 50000
true thrpt 4 31215.091 ± 671.106 ops/s
{noformat}
* Async profiler enabled (numbers basically reverse, patch looks a lot better
26k vs 23k)
{noformat}
main
Benchmark (nodeCount) (numReplicas) (numShards)
(preGenerate) (useSmallDocs) Mode Cnt Score Error Units
CloudIndexing.indexDoc 1 1 1
50000 true thrpt 4 22988.945 ± 1699.597 ops/s
patch
Benchmark (nodeCount) (numReplicas) (numShards)
(preGenerate) (useSmallDocs) Mode Cnt Score Error Units
CloudIndexing.indexDoc 1 1 1
50000 true thrpt 4 26276.781 ± 1308.604 ops/s
{noformat}
in short: without profiler enabled the change comes out a bit under main branch
BUT with profiler enabled there is a 15% boost in throughput.
I am still trying to wrap my head around why this is happening and the results
are not reproducible without profiling enabled
benchmark command for reference:
{noformat}
./jmh.sh -f 1 -r 300s -to 450s CloudIndexing -p nodeCount=1 -p numShards=1 -p
numReplicas=1 -jvmArgsAppend -Dsolr.http1=true
{noformat}
> UpdateLog refactor to use read/write locks
> ------------------------------------------
>
> Key: SOLR-17040
> URL: https://issues.apache.org/jira/browse/SOLR-17040
> Project: Solr
> Issue Type: Improvement
> Security Level: Public(Default Security Level. Issues are Public)
> Reporter: Alex Deparvu
> Assignee: Alex Deparvu
> Priority: Major
> Attachments: jfr-locks-main.png, jfr-locks-patch.png
>
> Time Spent: 10m
> Remaining Estimate: 0h
>
> Proposing a change to refactor the UpdateLog from all `synchronized (this)`
> blocks to read/write locks.
> Will post some benchmark results for evaluation if this is a useful change or
> not.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@solr.apache.org
For additional commands, e-mail: issues-h...@solr.apache.org
