[
https://issues.apache.org/jira/browse/HBASE-21590?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16720539#comment-16720539
]
Lars Hofhansl commented on HBASE-21590:
---------------------------------------
[~stack] In fact this triggers in many/most cases. Normally, most of the time
the top scanner on the Heap won't change (assuming data is compacted and/or
we're not updating individual KVs all the time - so that they end up in
different files). And if the next indexed key changes we need to reseek anyway.
[~zghaobac] NP at all. Thanks for looking and thanks for identifying the
problem in my initial solution in the first place! (Now at least we get some of
the performance back.)
> Optimize trySkipToNextColumn in StoreScanner a bit
> --------------------------------------------------
>
> Key: HBASE-21590
> URL: https://issues.apache.org/jira/browse/HBASE-21590
> Project: HBase
> Issue Type: Improvement
> Components: Performance, Scanners
> Reporter: Lars Hofhansl
> Assignee: Lars Hofhansl
> Priority: Critical
> Attachments: 21590-1.5.txt, HBASE-21590-master.txt
>
>
> See latest comment on HBASE-17958
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
