From: | Ragnar Kjørstad <postgres(at)ragnark(dot)vestdata(dot)no> |
---|---|
To: | Peter Darley <pdarley(at)kinesis-cem(dot)com> |
Cc: | Jeff Self <jself(at)nngov(dot)com>, Pgsql-Admin <pgsql-admin(at)postgresql(dot)org> |
Subject: | Re: Journaling file systems |
Date: | 2002-02-26 17:41:59 |
Message-ID: | 20020226184159.G13673@vestdata.no |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-admin |
On Tue, Feb 26, 2002 at 08:47:05AM -0800, Peter Darley wrote:
> > Friends,
> > I'm wondering if a journaling file system, like ext3, will make much
> > difference with the speed or reliability of PostgreSQL? Is it worth the
> > work to switch over?
PostgreSQL does not depend on journaling to be reliable.
> Is there some kind of file system (for Linux) that will provide better
> performance than ext2? XFS or ReiserFS maybe?
Both XFS and reiserfs are generally a little bit faster than ext2, put
primarely on file-operations (craete, delete, lookup). For raw data
transfer it's more a hardware issue than a filessytem issue (but XFS may
be a few percent faster).
Data-journaling could potentially improve performance a lot. By doing
writes to the journal istead of the actual file the number of seeks is
reduced and fsync can complete much faster. ext3 support data journaling
out of the box (mount-option) and there are experimental patches for
reiserfs.
I'm not sure how much seeking is going on without data-journaling? Maybe
writes are just appended to the log? could someone say more about this?
--
Ragnar Kjørstad
Big Storage
From | Date | Subject | |
---|---|---|---|
Next Message | Stephan Szabo | 2002-02-26 18:02:33 | Re: Size of database system |
Previous Message | Radu-Adrian Popescu | 2002-02-26 17:28:39 | Re: Journaling file systems |