[
https://issues.apache.org/jira/browse/CASSANDRA-1470?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12934452#action_12934452
]
Pavel Yaskevich commented on CASSANDRA-1470:
--------------------------------------------
Another possible option here without changing code base will be to decrease
number (to 10) in /proc/sys/vm/swappiness (default is 60), by doing that kernel
will prefer pruning page case and not swap. I was testing it on Linux (Intel(R)
Xeon(R) CPU E5430 @ 2.66GHz, 2 Gb memory) and it gave me a better read/write
results.
READ (final chunks):
(60 in /proc/sys/vm/swappiness)
97441,215,215,0.235058996032,536
99583,214,214,0.236029704857,546
100000,41,41,0.252929151201,549
(10 in /proc/sys/vm/swappiness)
97589,186,186,0.271465253625,486
99893,230,230,0.216397065773,496
100000,10,10,0.0753863036075,496
WRITE (final chunk):
(60 in /proc/sys/vm/swappiness)
91934,792,792,0.0656061102088,104
98274,634,634,0.0562482395759,114
100000,172,172,0.127704215216,118
(10 in /proc/sys/vm/swappiness)
85593,792,792,0.0618825450318,98
93518,792,792,0.0669648991473,109
100000,648,648,0.0500462510563,116
> use direct io for compaction
> ----------------------------
>
> Key: CASSANDRA-1470
> URL: https://issues.apache.org/jira/browse/CASSANDRA-1470
> Project: Cassandra
> Issue Type: Improvement
> Components: Core
> Reporter: Jonathan Ellis
> Assignee: Pavel Yaskevich
> Fix For: 0.7.1
>
> Attachments: 1470-v2.txt, 1470.txt, CASSANDRA-1470-for-0.6.patch,
> CASSANDRA-1470-v10-for-0.7.patch, CASSANDRA-1470-v11-for-0.7.patch,
> CASSANDRA-1470-v2.patch,
> CASSANDRA-1470-v3-0.7-with-LastErrorException-support.patch,
> CASSANDRA-1470-v4-for-0.7.patch, CASSANDRA-1470-v5-for-0.7.patch,
> CASSANDRA-1470-v6-for-0.7.patch, CASSANDRA-1470-v7-for-0.7.patch,
> CASSANDRA-1470-v8-for-0.7.patch, CASSANDRA-1470-v9-for-0.7.patch,
> CASSANDRA-1470.patch,
> use.DirectIORandomAccessFile.for.commitlog.against.1022235.patch
>
>
> When compaction scans through a group of sstables, it forces the data in the
> os buffer cache being used for hot reads, which can have a dramatic negative
> effect on performance.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
