stack commented on HBASE-20411:

Thanks [~huaxiang] ...  Let me put up what I have here (AtomicReference + 
create new instance of the data structure on every change -- gets rid of the 
synchronize). It looks too like a bunch of the memory accounting doesn't even 
have to be done in a thread safe manner so can do a form of what we did in 
TimeRangeTracker. Once up, lets come back around and look at your suggestion of 
letting the sizings run independent of each other. I think a hack-up patch 
first to see if stuff works if the sizings run independent would be first tack 
(If we could get away with it, your suggestion would be best.....).

> Ameliorate MutableSegment synchronize
> -------------------------------------
>                 Key: HBASE-20411
>                 URL: https://issues.apache.org/jira/browse/HBASE-20411
>             Project: HBase
>          Issue Type: Bug
>            Reporter: stack
>            Priority: Major
>         Attachments: 41901.lock.svg
> This item is migrated from HBASE-20236 so it gets dedicated issue.
> Let me upload evidence that has this synchronize as a stake in our write-time 
> perf. I'll migrate the patch I posted with updates that come of comments 
> posted by [~mdrob] on the HBASE-20236 issue.

This message was sent by Atlassian JIRA

Reply via email to