hbase-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Enis Soztutar (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HBASE-15643) Need metrics of cache hit ratio, etc for one table
Date Tue, 12 Jul 2016 21:06:20 GMT

    [ https://issues.apache.org/jira/browse/HBASE-15643?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15373693#comment-15373693
] 

Enis Soztutar commented on HBASE-15643:
---------------------------------------

Agreed. [~aliciashu] did some experiments with ycsb loading, and there was a perf hit about
3% of so. We were discussing offline about a way to carry the metrics reference in the HRegion
itself where it is initialized once per region lifecycle, and pass it down to the HFileBlockKey
to get rid of the map lookup. I think she will prototype the approach to see whether it works
or not. 

> Need metrics of cache hit ratio, etc for one table
> --------------------------------------------------
>
>                 Key: HBASE-15643
>                 URL: https://issues.apache.org/jira/browse/HBASE-15643
>             Project: HBase
>          Issue Type: Improvement
>            Reporter: Heng Chen
>            Assignee: Alicia Ying Shu
>         Attachments: HBASE-15643.patch
>
>
> There are many tables on our cluster,  only some of them need to be read online.  
> We could improve the performance of read by cache,  but we need some metrics for it at
table level. There are a few we can collect: BlockCacheCount, BlockCacheSize, BlockCacheHitCount,
BlockCacheMissCount, BlockCacheHitPercent



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message