Mailing-List: contact commits-help@cassandra.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@cassandra.apache.org
Date: Fri, 21 Nov 2014 14:43:35 +0000 (UTC)
From: "Alan Boudreault (JIRA)" <jira@apache.org>
To: commits@cassandra.apache.org
Message-ID: <JIRA.12755926.1416260622000.553995.1416581015016@Atlassian.JIRA>
In-Reply-To: <JIRA.12755926.1416260622000@Atlassian.JIRA>
References: <JIRA.12755926.1416260622000@Atlassian.JIRA>
 <JIRA.12755926.1416260622168@arcas>
Subject: [jira] [Comment Edited] (CASSANDRA-8329) LeveledCompactionStrategy
 should split large files across data directories when compacting
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit


    [ https://issues.apache.org/jira/browse/CASSANDRA-8329?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14220914#comment-14220914 ] 

Alan Boudreault edited comment on CASSANDRA-8329 at 11/21/14 2:43 PM:
----------------------------------------------------------------------

Devs, here my test results.

h4. Test

* 12 disks of 2G of size
* Cassandra use default values for concurrent_compactors and compaction_throughput_mb_per_sec.
* Goal: Stress the server with many concurent writes  for 45-50 minutes.

h5. Results - No Patch

We can see the peak the LCS compaction of big sstables.

!test_no_patch_2.0.jpg|thumbnaild!

h5. Results - With Patch

Success. There is no more peak during the compaction.

[^test_with_patch_2.0.jpg]

Let me know if I can do anything else.


was (Author: aboudreault):
Devs, here my test results.

h4. Test

* 12 disks of 2G of size
* Cassandra use default values for concurrent_compactors and compaction_throughput_mb_per_sec.
* Goal: Stress the server with many concurent writes  for 45-50 minutes.

h5. Results - No Patch

We can see the peak the LCS compaction of big sstables.

[^test_no_patch_2.0.jpg]

h5. Results - With Patch

Success. There is no more peak during the compaction.

[^test_with_patch_2.0.jpg]

Let me know if I can do anything else.


> LeveledCompactionStrategy should split large files across data directories when compacting
> ------------------------------------------------------------------------------------------
>
>                 Key: CASSANDRA-8329
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-8329
>             Project: Cassandra
>          Issue Type: Improvement
>          Components: Core
>            Reporter: J.B. Langston
>            Assignee: Marcus Eriksson
>             Fix For: 2.0.12
>
>         Attachments: 0001-get-new-sstable-directory-for-every-new-file-during-.patch, test_no_patch_2.0.jpg, test_with_patch_2.0.jpg
>
>
> Because we fall back to STCS for L0 when LCS gets behind, the sstables in L0 can get quite large during sustained periods of heavy writes.  This can result in large imbalances between data volumes when using JBOD support.  
> Eventually these large files get broken up as L0 sstables are moved up into higher levels; however, because LCS only chooses a single volume on which to write all of the sstables created during a single compaction, the imbalance is persisted.


--
This message was sent by Atlassian JIRA
(v6.3.4#6332)