You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@hbase.apache.org by "Hari Sekhon (JIRA)" <ji...@apache.org> on 2018/08/03 16:56:00 UTC
[jira] [Commented] (HBASE-18164) Much faster locality cost function
and candidate generator
[ https://issues.apache.org/jira/browse/HBASE-18164?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16568473#comment-16568473 ]
Hari Sekhon commented on HBASE-18164:
-------------------------------------
Has anyone tested if this results in faster convergence of optimum region placement for data locality after rolling restarts?
In enterprise HDP 2.6 still ships HBase 1.1.2 and I find it takes a very long time for data locality % to recover after dropping 30-40% after each rolling restart of the cluster.
When I say very long time, I mean days of gradually very slowly improving data locality (which I think is probably caused more by minor compactions than region migrations).
I've linked HBASE-21006 to cover poor locality after rolling restarts.
> Much faster locality cost function and candidate generator
> ----------------------------------------------------------
>
> Key: HBASE-18164
> URL: https://issues.apache.org/jira/browse/HBASE-18164
> Project: HBase
> Issue Type: Improvement
> Components: Balancer
> Reporter: Kahlil Oppenheimer
> Assignee: Kahlil Oppenheimer
> Priority: Critical
> Fix For: 1.4.0, 2.0.0-alpha-2, 2.0.0
>
> Attachments: 18164.branch-1.addendum.txt, HBASE-18164-00.patch, HBASE-18164-01.patch, HBASE-18164-02.patch, HBASE-18164-04.patch, HBASE-18164-05.patch, HBASE-18164-06.patch, HBASE-18164-07.patch, HBASE-18164-08.patch
>
>
> We noticed that during the stochastic load balancer was not scaling well with cluster size. That is to say that on our smaller clusters (~17 tables, ~12 region servers, ~5k regions), the balancer considers ~100,000 cluster configurations in 60s per balancer run, but only ~5,000 per 60s on our bigger clusters (~82 tables, ~160 region servers, ~13k regions) .
> Because of this, our bigger clusters are not able to converge on balance as quickly for things like table skew, region load, etc. because the balancer does not have enough time to "think".
> We have re-written the locality cost function to be incremental, meaning it only recomputes cost based on the most recent region move proposed by the balancer, rather than recomputing the cost across all regions/servers every iteration.
> Further, we also cache the locality of every region on every server at the beginning of the balancer's execution for both the LocalityBasedCostFunction and the LocalityCandidateGenerator to reference. This way, they need not collect all HDFS blocks of every region at each iteration of the balancer.
> The changes have been running in all 6 of our production clusters and all 4 QA clusters without issue. The speed improvements we noticed are massive. Our big clusters now consider 20x more cluster configurations.
> One design decision I made is to consider locality cost as the difference between the best locality that is possible given the current cluster state, and the currently measured locality. The old locality computation would measure the locality cost as the difference from the current locality and 100% locality, but this new computation instead takes the difference between the current locality for a given region and the best locality for that region in the cluster.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)