[
https://issues.apache.org/jira/browse/HBASE-16890?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15632371#comment-15632371
]
Duo Zhang commented on HBASE-16890:
-----------------------------------
And I have done some tests against an HDFS, with replication set to 1 and
disable low replication roll check.
{noformat}
./bin/hbase org.apache.hadoop.hbase.wal.WALPerformanceEvaluation -threads 100
-iterations 25000 -qualifiers 25 -keySize 50 -valueSize 200
FSHLog Summary: threads=100, iterations=25000, syncInterval=0 took 183.412s
13630.515ops/s
AsyncFSWAL Summary: threads=100, iterations=25000, syncInterval=0 took 113.391s
22047.605ops/s
AsyncFSWAL with my patch Summary: threads=100, iterations=25000, syncInterval=0
took 120.933s 20672.604ops/s
AsyncFSWAL with ram's patch Summary: threads=100, iterations=25000,
syncInterval=0 took 114.870s 21763.732ops/s
{noformat}
{noformat}
./bin/hbase org.apache.hadoop.hbase.wal.WALPerformanceEvaluation -threads 1
-iterations 250000 -qualifiers 25 -keySize 50 -valueSize 200
FSHLog Summary: threads=1, iterations=250000, syncInterval=0 took 100.482s
2488.008ops/s
AsyncFSWAL Summary: threads=1, iterations=250000, syncInterval=0 took 83.233s
3003.616ops/s
AsyncFSWAL with my patch Summary: Summary: threads=1, iterations=250000,
syncInterval=0 took 83.149s 3006.651ops/s
AsyncFSWAL with ram's patch test hang, can not finish
{noformat}
None of them can beat the original AsyncFSWAL implementation. I was wondering
why the results are different. Do you guys run all test instance in a single
node? NN, DN and WALPE on the same node so there is no network latency? I think
for most users, the HBsae will be deployed with normal network infrastructures
so the latency can not be ignored?
Thanks.
> Analyze the performance of AsyncWAL and fix the same
> ----------------------------------------------------
>
> Key: HBASE-16890
> URL: https://issues.apache.org/jira/browse/HBASE-16890
> Project: HBase
> Issue Type: Sub-task
> Components: wal
> Affects Versions: 2.0.0
> Reporter: ramkrishna.s.vasudevan
> Assignee: ramkrishna.s.vasudevan
> Fix For: 2.0.0
>
> Attachments: AsyncWAL_disruptor.patch, AsyncWAL_disruptor_1
> (2).patch, AsyncWAL_disruptor_3.patch, AsyncWAL_disruptor_3.patch,
> AsyncWAL_disruptor_4.patch, HBASE-16890-remove-contention-v1.patch,
> HBASE-16890-remove-contention.patch, Screen Shot 2016-10-25 at 7.34.47
> PM.png, Screen Shot 2016-10-25 at 7.39.07 PM.png, Screen Shot 2016-10-25 at
> 7.39.48 PM.png, async.svg, classic.svg, contention.png,
> contention_defaultWAL.png
>
>
> Tests reveal that AsyncWAL under load in single node cluster performs slower
> than the Default WAL. This task is to analyze and see if we could fix it.
> See some discussions in the tail of JIRA HBASE-15536.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
