[
https://issues.apache.org/jira/browse/IGNITE-11857?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16872422#comment-16872422
]
Aleksey Plekhanov commented on IGNITE-11857:
--------------------------------------------
I still can't reproduce a performance drop on my environment. Maximum I can get
is 1% drop on introduced for not MVCC transactions method
{{IgniteTxHandler#applyPartitionsUpdatesCounters}}.
So, I've tried to optimize {{PartitionTxUpdateCounterImpl#update(long, long)}}
and now this method about 1.3-1.5 times faster (according to microbenchmarks).
This method is synchronized. I've tried to get rid of the synchronized block
and preserve thread safety but this doesn't give any performance boost while
bringing extra complexity to the code. Seems like contention is not a problem
here.
Also, I've fixed some usages of write entries and read entries collections.
These collections are predicate views and {{F.isEmpty()}} on a collection leads
to undesirable iteration over collection sometimes.
But both these optimizations give no more than 0.5% on my environment.
[~ustas] can you check PR [1] against commit b87bea8 on your environment?
[~ascherbakov] cant you review partition update counter fix?
[1]: [https://github.com/apache/ignite/pull/6640]
> Investigate performance drop after IGNITE-10078
> -----------------------------------------------
>
> Key: IGNITE-11857
> URL: https://issues.apache.org/jira/browse/IGNITE-11857
> Project: Ignite
> Issue Type: Improvement
> Reporter: Alexei Scherbakov
> Assignee: Aleksey Plekhanov
> Priority: Major
> Attachments: ignite-config.xml,
> run.properties.tx-optimistic-put-b-backup
>
> Time Spent: 10m
> Remaining Estimate: 0h
>
> After IGNITE-10078 yardstick tests show performance drop up to 8% in some
> scenarios:
> * tx-optim-repRead-put-get
> * tx-optimistic-put
> * tx-putAll
> Partially this is due new update counter implementation, but not only.
> Investigation is required.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
