[
https://issues.apache.org/jira/browse/HBASE-11323?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14037749#comment-14037749
]
stack commented on HBASE-11323:
-------------------------------
Options are shaping up as follows (copied from HBASE-11364):
1. Do NOT enable offheap by default. Just talk it up as the way to go
underlining it will make pure in-memory access slower (but you can make it so
some of your tables are pegged in memory if you want because of the flag here).
Upside: No surprise. Downside: Folks don't read manuals nor change defaults.
2. Enable offheap BucketCache using CombinedBucketCache. When folks upgrade,
latency to user-level DATA blocks will go up. Upsides: less GC, more cached.
Downside: those who notice added latency might get upset. Changing schema will
require alter table.
3. Enable offheap BucketCache but in additive mode where we just add in an L2
under the L1 LruBlockCache. Upside: Additive. Downside: Could make for more GC.
Of the above, maybe 1. is the way to go? 2. may surprise in that perf and GC
gets better of a sudden (this would be ok) but others may be surprised that
their latencies have gone up for some key tables. 3. may actually make GC worse
(at least that is case in the SlabCache case which is similar and the L1/L2
layout doesn't get good review to date going by HBASE-8894).
Will test 3.
> BucketCache all the time!
> -------------------------
>
> Key: HBASE-11323
> URL: https://issues.apache.org/jira/browse/HBASE-11323
> Project: HBase
> Issue Type: Sub-task
> Components: io
> Reporter: stack
> Fix For: 0.99.0
>
> Attachments: ReportBlockCache.pdf
>
>
> One way to realize the parent issue is to just enable bucket cache all the
> time; i.e. always have offheap enabled. Would have to do some work to make
> it drop-dead simple on initial setup (I think it doable).
> So, upside would be the offheap upsides (less GC, less likely to go away and
> never come back because of full GC when heap is large, etc.).
> Downside is higher latency. In Nick's BlockCache 101 there is little to no
> difference between onheap and offheap. In a basic compare doing scans and
> gets -- details to follow -- I have BucketCache deploy about 20% less ops
> than LRUBC when all incache and maybe 10% less ops when falling out of cache.
> I can't tell difference in means and 95th and 99th are roughly same (more
> stable with BucketCache). GC profile is much better with BucketCache -- way
> less. BucketCache uses about 7% more user CPU.
> More detail on comparison to follow.
> I think the numbers disagree enough we should probably do the [~lhofhansl]
> suggestion, that we allow you to have a table sit in LRUBC, something the
> current bucket cache layout does not do.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
