[
https://issues.apache.org/jira/browse/HBASE-23296?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16974851#comment-16974851
]
Anoop Sam John commented on HBASE-23296:
----------------------------------------
I can say ur req in other words. You want to store the index/bloom blocks in
off heap BC and data blocks in file mode BC. Right now 2 issues which prevent
u from doing this
1. There can be only one kind of BC either off heap or file mode
2. BC will work iff LRU on heap cache (L1 its old name) is ON. But then also
the index/bloom blocks can sit *only* in L1 cache and data goes to BC.
I would say we should try solve this by looking at the req in above way.
We should be able to store the index/bloom blocks in BC also. I believe code
will handle it well even if the bloom/index data is backed by DBB. This should
work if or not the L1 cache is there or not.
There should be tiered BC.
> Support Bucket based L1 Cache
> -----------------------------
>
> Key: HBASE-23296
> URL: https://issues.apache.org/jira/browse/HBASE-23296
> Project: HBase
> Issue Type: Improvement
> Components: BlockCache
> Reporter: chenxu
> Priority: Major
>
> LruBlockCache is not suitable in the following scenarios:
> (1) cache size too large (will take too much heap memory, and
> evictBlocksByHfileName is not so efficient, as HBASE-23277 mentioned)
> (2) block evicted frequently, especially cacheOnWrite & prefetchOnOpen are
> enabled.
> Since block‘s data is reclaimed by GC, this may affect GC performance.
> So how about enabling a Bucket based L1 Cache.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
