hbase-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Ted Yu (JIRA)" <j...@apache.org>
Subject [jira] [Comment Edited] (HBASE-15065) SimpleRegionNormalizer should return multiple normalization plans in one run
Date Thu, 07 Jan 2016 02:03:39 GMT

    [ https://issues.apache.org/jira/browse/HBASE-15065?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15086422#comment-15086422
] 

Ted Yu edited comment on HBASE-15065 at 1/7/16 2:02 AM:
--------------------------------------------------------

Thanks for detailed comment.

bq. with this logic we merge 2 out of 3 10-size regions and split 60

After merging 2 of the size 10 regions, average becomes 90 / 3 = 30. So both the merge and
split are still fine.
If size 60 region gets split first, average becomes 90 / 5 = 18. The merging doesn't seem
necessary. However, size 60 region comes last. Meaning, potentially the normalization process
may have O(N^2) complexity where N is the number of regions in the table.

I think we don't need to compute new average for each iteration.

bq. we can run them concurrently more than 1 at a time

The splitting / merging request is asynchronous. So by executing multiple plans in one round,
the above is achieved.


was (Author: yuzhihong@gmail.com):
Thanks for detailed comment.

bq. with this logic we merge 2 out of 3 10-size regions and split 60

After merging 2 of the size 10 regions, average becomes 90 / 3 = 30. So both the merge and
split are still fine.
If size 60 region gets split first, average becomes 90 / 5 = 18. The merging doesn't seem
necessary. However, size 60 region comes last. Meaning, potentially the normalization process
may have O(N) complexity where N is the number of regions in the table.

I think we don't need to compute new average for each iteration.

bq. we can run them concurrently more than 1 at a time

The splitting / merging request is asynchronous. So by executing multiple plans in one round,
the above is achieved.

> SimpleRegionNormalizer should return multiple normalization plans in one run
> ----------------------------------------------------------------------------
>
>                 Key: HBASE-15065
>                 URL: https://issues.apache.org/jira/browse/HBASE-15065
>             Project: HBase
>          Issue Type: Bug
>            Reporter: Ted Yu
>            Assignee: Ted Yu
>         Attachments: 15065-v1.txt
>
>
> This is follow up to HBASE-14867 w.r.t. SimpleRegionNormalizer
> Outline for enhancements:
> * adjustment to the period when SimpleRegionNormalizer runs
> * explore merge opportunities among all neighboring region pairs
> * return multiple normalization plans



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

Mime
View raw message