[ https://issues.apache.org/jira/browse/HBASE-14463?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14966438#comment-14966438 ]
Yu Li commented on HBASE-14463: ------------------------------- Thanks all for taking a look here. Was trying to reproduce [~anoop.hbase]/[~ram_krish]'s result and do some investigation but met with some problem, such as jvm crash during data ingestion with PE (haven't file any JIRA since not sure whether it's an env-specific issue) and AssertionError during multi get testing (see HBASE-14660). Now I could get the test run after disabling assertion and will do further debugging, will update my findings later. [~jingcheng...@intel.com] I also doubt about the purge call slows down the performance, will add some threshold there and check the perf comparison. Thanks for point it out. [~lhofhansl] we need to store the lock(entry) somewhere and using lockPool is for reducing lock contention. I think the idea of using weak reference is good but lack of some perf testing here before. Or any better idea please let me know :-) > 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, 0.98.16 > > Attachments: GC_with_WeakObjectPool.png, HBASE-14463.patch, > HBASE-14463_v11.patch, HBASE-14463_v12.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-latest.png, TestBucketCache_with_IdLock.png, > TestBucketCache_with_IdReadWriteLock-latest.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)