You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@cassandra.apache.org by "Ariel Weisberg (JIRA)" <ji...@apache.org> on 2015/01/15 23:32:34 UTC
[jira] [Updated] (CASSANDRA-8634) Multi-thread commit log
compression
[ https://issues.apache.org/jira/browse/CASSANDRA-8634?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Ariel Weisberg updated CASSANDRA-8634:
--------------------------------------
Description:
The initial commit log compression implementation can do about 240 megabytes/second (uncompressed input) on a fast (500 megabytes/sec) SSD.
This is basically what I measured as the throughput of LZ4 against the same data.
Assuming a 1.5x compression ratio that is 160 megabytes/second of output. Enough to saturate a spinning disk, but not an SSD.
was:
The initial commit log compression implementation can do about 240 megabytes/second (uncompressed input) on a fast (500 megabytes/sec) SSD.
Assuming a 1.5x compression ratio that is 160 megabytes/second of output. Enough to saturate a spinning disk, but not an SSD.
> Multi-thread commit log compression
> -----------------------------------
>
> Key: CASSANDRA-8634
> URL: https://issues.apache.org/jira/browse/CASSANDRA-8634
> Project: Cassandra
> Issue Type: Improvement
> Components: Core
> Reporter: Ariel Weisberg
> Labels: performance
>
> The initial commit log compression implementation can do about 240 megabytes/second (uncompressed input) on a fast (500 megabytes/sec) SSD.
> This is basically what I measured as the throughput of LZ4 against the same data.
> Assuming a 1.5x compression ratio that is 160 megabytes/second of output. Enough to saturate a spinning disk, but not an SSD.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)