hbase-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Lars Hofhansl (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HBASE-13103) [ergonomics] add region size balancing as a feature of master
Date Thu, 09 Apr 2015 06:47:15 GMT

    [ https://issues.apache.org/jira/browse/HBASE-13103?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14486841#comment-14486841
] 

Lars Hofhansl commented on HBASE-13103:
---------------------------------------

Yeah, meant the "reshaping" after I identified that something is odd/bad about a table.
But maybe it's better to just automate, otherwise nobody would use it, as you say.

Splits already happen automatically with nice simple local-only logic do we need more logic
for those? (but we could get rid of IncreasingToUpperBoundRegionSplitPolicy and combine it
all in one class, which would be nice).

bq. could have policy which is much more conservative in merging, than it is in splitting
I think that'd be nice. With IncreasingToUpperBoundRegionSplitPolicy it's possible that we
get a 2x size difference between regions for a bit. Hard to say whether a region will be written
to in the future, and avoid an early merge. Maybe we can track the age of a region? And then
favor older regions for merges unless they're hot...

bq. ServerLoad and RegionLoad won't suffice you think?
You're right, that's probably all the information we need. And if not, we'd add it.


> [ergonomics] add region size balancing as a feature of master
> -------------------------------------------------------------
>
>                 Key: HBASE-13103
>                 URL: https://issues.apache.org/jira/browse/HBASE-13103
>             Project: HBase
>          Issue Type: Brainstorming
>          Components: Usability
>            Reporter: Nick Dimiduk
>            Assignee: Mikhail Antonov
>             Fix For: 2.0.0, 1.1.0
>
>         Attachments: HBASE-13103-v0.patch
>
>
> Often enough, folks miss-judge split points or otherwise end up with a suboptimal number
of regions. We should have an automated, reliable way to "reshape" or "balance" a table's
region boundaries. This would be for tables that contain existing data. This might look like:
> {noformat}
> Admin#reshapeTable(TableName, int numSplits);
> {noformat}
> or from the shell:
> {noformat}
> > reshape TABLE, numSplits
> {noformat}
> Better still would be to have a maintenance process, similar to the existing Balancer
that runs AssignmentManager on an interval, to run the above "reshape" operation on an interval.
That way, the cluster will automatically self-correct toward a desirable state.



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

Mime
View raw message