[
https://issues.apache.org/jira/browse/HBASE-26067?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17450324#comment-17450324
]
Wellington Chevreuil edited comment on HBASE-26067 at 11/29/21, 10:21 AM:
--------------------------------------------------------------------------
{quote}
So we could have about 6% performance gain in YCSB? It is a bit surprise to me,
I didn't think we could increase read/write performance, we could just save
lots of time for the background tasks, such as flush and compaction...{quote}
I think part of it is due to split storms happening during the run. Splits were
taking longer to complete on the hboss/renames cluster, causing some requests
to timeout for longer. Another observation was that the hboss/renames cluster
had more regions at the end of the run. Maybe because compactions were much
slower, the split threshold was being met more often, leading to more splits.
Snapshots recovery is another area we could see improvements, when dealing with
snapshots of tables with hundreds to thousands of regions. We are doing more
tests to consolidate numbers, will post more details here later.
was (Author: wchevreuil):
{quote}
So we could have about 6% performance gain in YCSB? It is a bit surprise to me,
I didn't think we could increase read/write performance, we could just save
lots of time for the background tasks, such as flush and compaction...
{quote}
I think part of it is due to split storms happening during the run. Splits were
taking longer to complete on the hboss/renames cluster, causing some requests
to timeout for longer. Another observation was that the hboss/renames cluster
had more regions at the end of the run. Maybe because compactions were much
slower, the split threshold was being met more often, leading to more splits.
Snapshots recovery is another area we could see improvements, when dealing with
snapshots of tables with hundreds to thousands of regions. We are doing more
tests to consolidate numbers, will post more details here later.
> Change the way on how we track store file list
> ----------------------------------------------
>
> Key: HBASE-26067
> URL: https://issues.apache.org/jira/browse/HBASE-26067
> Project: HBase
> Issue Type: Umbrella
> Components: HFile
> Reporter: Duo Zhang
> Assignee: Duo Zhang
> Priority: Major
>
> Open a separated jira to track the work since it can not be fully included in
> HBASE-24749.
> I think this could be a landed prior to HBASE-24749, as if this works, we
> could have different implementations for tracking store file list.
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
