[
https://issues.apache.org/jira/browse/HBASE-23887?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17128972#comment-17128972
]
Bharath Vissapragada commented on HBASE-23887:
----------------------------------------------
Thanks, mind putting it in a Google doc and opening it up comments to public? I
think I got a sense of what you are doing. Added some more comments in the
patch.
(TL;DR) The cache becomes some what adaptive depending on the pattern of
evictions and how much memory is freed in each eviction. With the new patch,
the % of blocks we skip when caching dynamically changes depending on all the
knobs mentioned above. What blocks are cached is still random (which the other
reviewers already pointed out in the comments). The idea is definitely
interesting (as depicted in the benchmarks), since it avoids thrashing and
un-necessary GCs in some bad workloads. The most critical part IMO is how the
cache adapts to changing workloads, basically avoiding cache misses once the
workload changes to a non-scan type.
While looking at the patch, I was wondering if we should implement this as a
separate Cache implementation that extends LruBlockCache, say
AdaptiveLruBlockCache or something? (restructure the code to override certain
methods like evict() etc). I think that helps with the following
1. Since LruBlockCache is heavily used, subclassing it will prevent any new
bugs and no scope of regressions during upgrade.
2. The patch adds a bunch of new parameters and there is a lot of math around
them. All of the code will be consolidated nicely in this class. Easier to read.
What do others think?
> BlockCache performance improve by reduce eviction rate
> ------------------------------------------------------
>
> Key: HBASE-23887
> URL: https://issues.apache.org/jira/browse/HBASE-23887
> Project: HBase
> Issue Type: Improvement
> Components: BlockCache, Performance
> Reporter: Danil Lipovoy
> Assignee: Danil Lipovoy
> Priority: Minor
> Attachments: 1582787018434_rs_metrics.jpg,
> 1582801838065_rs_metrics_new.png, BC_LongRun.png,
> BlockCacheEvictionProcess.gif, cmp.png, evict_BC100_vs_BC23.png,
> eviction_100p.png, eviction_100p.png, eviction_100p.png, gc_100p.png,
> graph.png, image-2020-06-07-08-11-11-929.png,
> image-2020-06-07-08-19-00-922.png, image-2020-06-07-12-07-24-903.png,
> image-2020-06-07-12-07-30-307.png, image-2020-06-08-17-38-45-159.png,
> image-2020-06-08-17-38-52-579.png, image-2020-06-08-18-35-48-366.png,
> read_requests_100pBC_vs_23pBC.png, requests_100p.png, requests_100p.png,
> requests_new2_100p.png, requests_new_100p.png, scan.png, wave.png
>
>
> Hi!
> I first time here, correct me please if something wrong.
> I want propose how to improve performance when data in HFiles much more than
> BlockChache (usual story in BigData). The idea - caching only part of DATA
> blocks. It is good becouse LruBlockCache starts to work and save huge amount
> of GC.
> Sometimes we have more data than can fit into BlockCache and it is cause a
> high rate of evictions. In this case we can skip cache a block N and insted
> cache the N+1th block. Anyway we would evict N block quite soon and that why
> that skipping good for performance.
> Example:
> Imagine we have little cache, just can fit only 1 block and we are trying to
> read 3 blocks with offsets:
> 124
> 198
> 223
> Current way - we put the block 124, then put 198, evict 124, put 223, evict
> 198. A lot of work (5 actions).
> With the feature - last few digits evenly distributed from 0 to 99. When we
> divide by modulus we got:
> 124 -> 24
> 198 -> 98
> 223 -> 23
> It helps to sort them. Some part, for example below 50 (if we set
> *hbase.lru.cache.data.block.percent* = 50) go into the cache. And skip
> others. It means we will not try to handle the block 198 and save CPU for
> other job. In the result - we put block 124, then put 223, evict 124 (3
> actions).
> See the picture in attachment with test below. Requests per second is higher,
> GC is lower.
>
> The key point of the code:
> Added the parameter: *hbase.lru.cache.data.block.percent* which by default =
> 100
>
> But if we set it 1-99, then will work the next logic:
>
>
> {code:java}
> public void cacheBlock(BlockCacheKey cacheKey, Cacheable buf, boolean
> inMemory) {
> if (cacheDataBlockPercent != 100 && buf.getBlockType().isData())
> if (cacheKey.getOffset() % 100 >= cacheDataBlockPercent)
> return;
> ...
> // the same code as usual
> }
> {code}
>
> Other parameters help to control when this logic will be enabled. It means it
> will work only while heavy reading going on.
> hbase.lru.cache.heavy.eviction.count.limit - set how many times have to run
> eviction process that start to avoid of putting data to BlockCache
> hbase.lru.cache.heavy.eviction.bytes.size.limit - set how many bytes have to
> evicted each time that start to avoid of putting data to BlockCache
> By default: if 10 times (100 secunds) evicted more than 10 MB (each time)
> then we start to skip 50% of data blocks.
> When heavy evitions process end then new logic off and will put into
> BlockCache all blocks again.
>
> Descriptions of the test:
> 4 nodes E5-2698 v4 @ 2.20GHz, 700 Gb Mem.
> 4 RegionServers
> 4 tables by 64 regions by 1.88 Gb data in each = 600 Gb total (only FAST_DIFF)
> Total BlockCache Size = 48 Gb (8 % of data in HFiles)
> Random read in 20 threads
>
> I am going to make Pull Request, hope it is right way to make some
> contribution in this cool product.
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
