[
https://issues.apache.org/jira/browse/HBASE-14463?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14910019#comment-14910019
]
Yu Li commented on HBASE-14463:
-------------------------------
Hi [~ikeda],
Yes I totally understand your concern, that one block eviction will cause
blocking on reading another unrelated block. However, as I mentioned before,
the blockcache should be a *one write many read* thing, or say much more hits
than eviction (if not the case, I think it's recommended to disable blockcache
to save the miss and load cost). So I think the cost is acceptable, and the
test result with block-evicting thread could prove my statement.
OTOH, I agree that we could do lock striping for further improvement, but per
the above comment I think it's not that critical w/o such improvements, and I'd
prefer to open another JIRA for it instead of blocking this one (Rome was not
built in a day, right :-)). Makes sense?
[~stack], [~anoop.hbase], [~tedyu] and [[email protected]], please also
let me know your thoughts, thanks.
> Severe performance downgrade when parallel reading a single key from
> BucketCache
> --------------------------------------------------------------------------------
>
> Key: HBASE-14463
> URL: https://issues.apache.org/jira/browse/HBASE-14463
> Project: HBase
> Issue Type: Bug
> Affects Versions: 0.98.14, 1.1.2
> Reporter: Yu Li
> Assignee: Yu Li
> Fix For: 2.0.0, 1.2.0, 1.3.0, 1.0.3, 1.1.3, 0.98.16
>
> Attachments: HBASE-14463.patch, HBASE-14463_v2.patch,
> HBASE-14463_v3.patch, HBASE-14463_v4.patch, HBASE-14463_v5.patch,
> TestBucketCache-new_with_IdLock.png,
> TestBucketCache-new_with_IdReadWriteLock.png,
> TestBucketCache_with_IdLock.png,
> TestBucketCache_with_IdReadWriteLock-resolveLockLeak.png,
> TestBucketCache_with_IdReadWriteLock.png
>
>
> We store feature data of online items in HBase, do machine learning on these
> features, and supply the outputs to our online search engine. In such
> scenario we will launch hundreds of yarn workers and each worker will read
> all features of one item(i.e. single rowkey in HBase), so there'll be heavy
> parallel reading on a single rowkey.
> We were using LruCache but start to try BucketCache recently to resolve gc
> issue, and just as titled we have observed severe performance downgrade.
> After some analytics we found the root cause is the lock in
> BucketCache#getBlock, as shown below
> {code}
> try {
> lockEntry = offsetLock.getLockEntry(bucketEntry.offset());
> // ...
> if (bucketEntry.equals(backingMap.get(key))) {
> // ...
> int len = bucketEntry.getLength();
> Cacheable cachedBlock = ioEngine.read(bucketEntry.offset(), len,
> bucketEntry.deserializerReference(this.deserialiserMap));
> {code}
> Since ioEnging.read involves array copy, it's much more time-costed than the
> operation in LruCache. And since we're using synchronized in
> IdLock#getLockEntry, parallel read dropping on the same bucket would be
> executed in serial, which causes a really bad performance.
> To resolve the problem, we propose to use ReentranceReadWriteLock in
> BucketCache, and introduce a new class called IdReadWriteLock to implement it.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
