[
https://issues.apache.org/jira/browse/HBASE-26590?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17469506#comment-17469506
]
Huaxiang Sun commented on HBASE-26590:
--------------------------------------
I modified my testing case, excluding connection setup/teardown from the time
counted. Here is the result for 1m random meta lookup. I added option to use
BlockingRpcClient for meta lookup against the default NettyRpcClient.
||h5. ~Version~ ||h5. ~Meta Replica Load Balance Enabled~||h5.
~BlockingRpcClient~ ||h5. ~Time(ms)~||
||h5. ~2.4.5-with-fixed~||h5. ~No~||h5. ~No~||h5. ~370814~||
||h5. ~2.4.5-with-fixed~||h5. ~No~||h5. ~Yes~||h5. ~358931~||
||h5. ~2.4.5-with-fixed~||h5. ~Yes~||h5. ~Yes~||h5. ~349485~ ||
||h5. ~2.4.5~||h5. ~No~||h5. ~No~||h5. ~516640~ ||
||h5. ~2.4.5~||h5. ~Yes~||h5. ~Yes~||h5. ~497509~||
||h5. ~cdh-5.16.2~||h5. ~No~||h5. ~No~||h5. ~371540~||
When I did the Table.get() test. It is hard to draw a solid conclusion due to
key distribution, most of the keys randomly created fall into the the last
region and it is cached. BlockingRpcClient/NettyRpcClient difference is about
3% (Not as initially reported as 5 ~ 10%), so not a very big concern here.
This difference here is not big as what we observed at the production cluster.
I am going to put up the patch and will work with the team to see if it helps.
> Hbase-client Meta lookup performance regression between hbase-1 and hbase-2
> ---------------------------------------------------------------------------
>
> Key: HBASE-26590
> URL: https://issues.apache.org/jira/browse/HBASE-26590
> Project: HBase
> Issue Type: Improvement
> Components: meta
> Affects Versions: 2.4.0, 2.5.0, 2.3.7, 2.6.0
> Reporter: Huaxiang Sun
> Assignee: Huaxiang Sun
> Priority: Major
>
> One of our users complained higher latency after application upgrades from
> hbase-1.2 client (CDH-5.16.2) to hbase-2.4.5 client with meta replica Load
> Balance mode during app restart. I reproduced the regression by a test for
> meta lookup.
> At my test cluster, there are 160k regions for the test table, so there are
> 160k entries in meta region. Used one thread to do 1 million meta lookup
> against the meta region server.
>
> ||Version ||Meta Replica Load Balance Enabled||Time ||
> ||2.4.5-with-fixed||Yes||336458ms||
> ||2.4.5-with-fixed||No||333253ms||
> ||2.4.5||Yes||469980ms||
> ||2.4.5||No||470515ms||
> | *cdh-5.16.2*| *No* | *323412ms*|
>
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
