[ 
https://issues.apache.org/jira/browse/HBASE-22285?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17111790#comment-17111790
 ] 

Aman Poonia commented on HBASE-22285:
-------------------------------------

[~ndimiduk]- Yes sir your review is correct. The reason to introduce new 
Normalizer is because i wanted to keep the things simpler. A 
simpleRegionNormalizer does multiple things. It split the region and it merges 
the region depending on the average region size and there was no configuratin 
to do only merge or only splits.



What i wanted to achieve was just merge the regions to make sure we don't have 
zero bytes or very small region(may be less than a MB or GB). Adding the 
configuration helps in achieving the same thing but then we will have too many 
configuration for one normalizer and predicting it's behaviour becomes bit 
difficult. So for the sake of simplicity i created a new Normalizer and 
abstracted the common functionality out.

If you think adding configuration is a good way then adding a new class then i 
am fine with that also. Have no reservation for this approach except above 
mentioned reasons.

> A normalizer which merges very small size regions with adjacent 
> regions.(MergeToNormalize)
> ------------------------------------------------------------------------------------------
>
>                 Key: HBASE-22285
>                 URL: https://issues.apache.org/jira/browse/HBASE-22285
>             Project: HBase
>          Issue Type: New Feature
>          Components: master
>            Reporter: Aman Poonia
>            Assignee: Aman Poonia
>            Priority: Minor
>             Fix For: 3.0.0-alpha-1, 2.3.0
>
>
> There are scenarios where we have seen around 5% of total regions with a size 
> of 0 bytes and another 5-6 % regions with size in a few bytes. These kinds of 
> regions increase with time considering we have TTL over the rows. 
> After exploring the option of RegionNormalizer and doing some quick runs we 
> found that that is not suitable considering it also splits the regions and 
> merges to normalize. What we really want is to split as per Split policy and 
> merge very small regions with adjacent regions to make sure we reduce 0-byte 
> regions.
> We can plugin this normalizer using the property 
> "hbase.master.normalizer.class"



--
This message was sent by Atlassian Jira
(v8.3.4#803005)

Reply via email to