[
https://issues.apache.org/jira/browse/HBASE-17081?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15811428#comment-15811428
]
Anastasia Braginsky commented on HBASE-17081:
---------------------------------------------
Hey Everybody,
I was out for about ten days due to hard personal reasons, I wasn’t connected
to what was going on here with the compacting pipeline synchronization and etc.
Now I am back to business.
I am not yet familiar with all the details, but I read all the JIRAs briefly
and I have a sense of what is going on.
I see that HBASE-17081 was taken out again and I think that again it was wrong
decision to do.
The problem of synchronization of getScanners() in CompactionPipeline indeed
exist, but 17081 just opened a bigger problem of synch issues in
CompactionPipeline which is now being fixed under HBASE-17434.
I want to stress that *we need HBASE-17081 in!*
We have seen *tremendous performance difference* in the performance, when we
working with and without CompositeSnapshot!
This is the game changer!
The nice graphs you have seen in the attached presentation (50% improvement in
the median in the read latency(!)) are so nice only when with composite
snapshot.
The two times this commit was reverted the revert could be avoided.
But anyway we are where we are and I wasn’t there on time to defend 17081.
Anyway, currently I do not want to compete with re-bases and concurrent changes
of the Compaction Pipeline done under HBASE-17434.
Currently I am taking my hands of composite snapshot and I will try to return
after HBASE-17434 will be in.
No need to take it into any separate branch.
> Flush the entire CompactingMemStore content to disk
> ---------------------------------------------------
>
> Key: HBASE-17081
> URL: https://issues.apache.org/jira/browse/HBASE-17081
> Project: HBase
> Issue Type: Sub-task
> Reporter: Anastasia Braginsky
> Assignee: Eshcar Hillel
> Fix For: 2.0.0
>
> Attachments: HBASE-15787_8.patch, HBASE-17081-V01.patch,
> HBASE-17081-V02.patch, HBASE-17081-V03.patch, HBASE-17081-V04.patch,
> HBASE-17081-V05.patch, HBASE-17081-V06.patch, HBASE-17081-V06.patch,
> HBASE-17081-V07.patch, HBASE-17081-V10.patch,
> HBaseMeetupDecember2016-V02.pptx, Pipelinememstore_fortrunk_3.patch
>
>
> Part of CompactingMemStore's memory is held by an active segment, and another
> part is divided between immutable segments in the compacting pipeline. Upon
> flush-to-disk request we want to flush all of it to disk, in contrast to
> flushing only tail of the compacting pipeline.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
