[ 
https://issues.apache.org/jira/browse/HBASE-25882?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17400370#comment-17400370
 ] 

Xiaolin Ha commented on HBASE-25882:
------------------------------------

[~claraxiong] [~zhangduo] please take a look, I updated the description of the 
issue. And I don't think make the multipler of TableSkewCostFunction to 0 in 
this circumstance is a good method, because it means when I want to disable the 
`hbase.master.loadbalance.bytable`, I also need to set the multipler of 
TableSkewCostFunction back to more than 0. Obviously, it's not convenient and 
is easy to forget. Thanks.

> TableSkewCostFunction may cost unnecessary calculation steps when balancing 
> by table
> ------------------------------------------------------------------------------------
>
>                 Key: HBASE-25882
>                 URL: https://issues.apache.org/jira/browse/HBASE-25882
>             Project: HBase
>          Issue Type: Improvement
>          Components: Balancer
>    Affects Versions: 3.0.0-alpha-1, 2.0.0
>            Reporter: Xiaolin Ha
>            Assignee: Xiaolin Ha
>            Priority: Major
>         Attachments: image-2021-08-17-20-20-12-768.png
>
>
> When using balance by table, the StochasticLoadBalancer will create the 
> cluster state according to the regions distribution of only one table. As a 
> result,  the TableSkewCostFunction should be replaced by the 
> RegionCountSkewCostFunction when the table count of the cluster state is less 
> than 2.
> The most important problem is that,  TableSkewCostFunction will cause 
> unnecessary calculation steps when there is only one table.
> For example,we have 40000+ regions for one table in one group, balance this 
> table may calculate lots of steps, but we can avoid duplicate computing costs 
> of TableSkewCostFunction while computing cost of RegionCountSkewCostFunction 
> and balancing by table.
>  
>   !image-2021-08-17-20-20-12-768.png|width=907,height=164!



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

Reply via email to