Re: Filesystems comparison using sql-bench and mySQL
Sonny Rao <[email protected]> Tue, 19 Apr 2005 15:36:25 -0400
| Newsgroups | gmane.comp.file-systems.ext2.devel,gmane.comp.db.mysql.benchmarks |
|---|---|
| Message-ID | <[email protected]> |
On Tue, Apr 19, 2005 at 06:08:00PM +0200, Laurent Vivier wrote: > Hi, > > I made new measurements according comments from Peter. > In this case, the table type is InnoDB. This type generates a lot of > I/O. > > Results are here: > > http://www.bullopensource.org/ext4/sqlbench/InnoDB/ > > Conclusion: > > - Avoid JFS and XFS. > - Use ReiserFS > - but patching ext3 could improve performance (compared to ext3 alone) > > Regards, > Laurent This is very interesting, because we've only seen JFS (and sometimes XFS) perform poorly when you're doing a lot of metadata operations due to problems in the log handling code. Can we get iostat, oprofile, and vmstat data for these runs ? Sonny ------------------------------------------------------- This SF.Net email is sponsored by: New Crystal Reports XI. Version 11 adds new functionality designed to reduce time involved in creating, integrating, and deploying reporting solutions. Free runtime info, new features, or free trial, at: http://www.businessobjects.com/devxi/728