[
https://issues.apache.org/jira/browse/HBASE-16890?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15602833#comment-15602833
]
stack commented on HBASE-16890:
-------------------------------
Hang on.... Rerunning. I had TRACE level logging going on for DFSClient...
thats not fair in compare. Sorry [~ram_krish] On rerun I see this:
{code}
5610 Performance counter stats for './hbase/bin/hbase --config
/home/stack/conf_hbase org.apache.hadoop.hbase.wal.WALPerformanceEvaluation
-threads 100 -iterat ions 250000 -keySize 50 -valueSize 100 -qualifiers 10':
5611
5612 4036698.163447 task-clock (msec) # 9.891 CPUs utilized
5613 41,344,568 context-switches # 0.010 M/sec
5614 6,909,262 cpu-migrations # 0.002 M/sec
5615 12,455,604 page-faults # 0.003 M/sec
5616 7,548,061,036,706 cycles # 1.870 GHz
5617 <not supported> stalled-cycles-frontend
5618 <not supported> stalled-cycles-backend
5619 2,788,283,330,940 instructions # 0.37 insns per cycle
5620 480,911,981,821 branches # 119.135 M/sec
5621 5,351,642,769 branch-misses # 1.11% of all branches
5622
5623 408.109482919 seconds time elapsed
{code}
So, asyncwal is 30% slower than DFSClient#DFSOS
> 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: 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)
