[ 
https://issues.apache.org/jira/browse/HBASE-17204?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16287088#comment-16287088
 ] 

Anoop Sam John commented on HBASE-17204:
----------------------------------------

Thanks Stack...     Ya we dont have even a way as of now , to have only BC and 
NO on heap LRU cache at all.  All bloom, index , kind of meta blocks go to LRU.
Only diff will be this. Previously even the data blocks from the META table and 
Namespace table used to go to LRU cache.  But now that also considered as data 
blocks and go to BC. And BC by default is off heap. And off heap BC perf is 
even better than on heap (Specially when u have bigger sized cache needs and so 
java heap).  But when it is in File mode BC, ya the perf will be lesser. But 
META data info (region locations) are cached at client end connection side. 
Also if there are more looks up to META table (region moves or splits) from 
different connections, the file system cache (OS cache) also might help us.  So 
not any major concern. Still I have a task to specifically test the META table 
only read workload to compare on heap LRU vs File mode BC.  Will do this week 
and let u know.. Just explaining all the changes so that u can better predict 
any possible impact and help to correct if any needed.  Tks Zach.

> Make Off heap Bucket Cache default ON
> -------------------------------------
>
>                 Key: HBASE-17204
>                 URL: https://issues.apache.org/jira/browse/HBASE-17204
>             Project: HBase
>          Issue Type: Improvement
>    Affects Versions: 2.0.0
>            Reporter: Anoop Sam John
>            Assignee: Anoop Sam John
>            Priority: Blocker
>             Fix For: 2.0.0-beta-1
>
>
> L2 cache can be used for data blocks.  By default it is off now. After 
> HBASE-11425 work, L2 off heap cache can equally perform with L1 on heap 
> cache. On heavy loaded workload, this can even out perform L1 cache.  Pls see 
> recently published report by Alibaba.  Also this work was backported by 
> Rocketfuel and similar perf improvement report from them too.
> Let us turn L2 off heap cache ON. As it is off heap, we can have much larger 
> sized L2 BC.  What should be the default size?



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)

Reply via email to