From: Mark Fasheh <mark.fasheh@oracle.com>
To: Andrew Morton <akpm@osdl.org>
Cc: Daniel Phillips <phillips@google.com>,
linux-kernel@vger.kernel.org, ocfs2-devel@oss.oracle.com
Subject: [Ocfs2-devel] Ocfs2 performance bugs of doom
Date: Sun, 5 Mar 2006 11:22:48 -0800 [thread overview]
Message-ID: <20060305192247.GH31587@ca-server1.us.oracle.com> (raw)
In-Reply-To: <20060303233617.51718c8e.akpm@osdl.org>
On Fri, Mar 03, 2006 at 11:36:17PM -0800, Andrew Morton wrote:
> > - hash = full_name_hash(name, len);
>
> err, you might want to calculate that hash outside the spinlock.
Hmm, good catch.
> Maybe have a lock per bucket, too.
That could be an avenue to explore. I guess we probably want to optimize the
actual lookup and memory usage before looking at that.
> A 1MB hashtable is verging on comical. How may data are there in total?
Yes 1MB is much too big. We're looking right now at 3 lock resources per
inode, so total number of elements depends upon your file system I suppose.
Hopefully I'll have some time today to run some basic tests which may show
us what size strikes a better balance of performance versus memory overhead.
--Mark
--
Mark Fasheh
Senior Software Developer, Oracle
mark.fasheh at oracle.com
WARNING: multiple messages have this Message-ID (diff)
From: Mark Fasheh <mark.fasheh@oracle.com>
To: Andrew Morton <akpm@osdl.org>
Cc: Daniel Phillips <phillips@google.com>,
linux-kernel@vger.kernel.org, ocfs2-devel@oss.oracle.com
Subject: Re: Ocfs2 performance bugs of doom
Date: Sun, 5 Mar 2006 11:22:48 -0800 [thread overview]
Message-ID: <20060305192247.GH31587@ca-server1.us.oracle.com> (raw)
In-Reply-To: <20060303233617.51718c8e.akpm@osdl.org>
On Fri, Mar 03, 2006 at 11:36:17PM -0800, Andrew Morton wrote:
> > - hash = full_name_hash(name, len);
>
> err, you might want to calculate that hash outside the spinlock.
Hmm, good catch.
> Maybe have a lock per bucket, too.
That could be an avenue to explore. I guess we probably want to optimize the
actual lookup and memory usage before looking at that.
> A 1MB hashtable is verging on comical. How may data are there in total?
Yes 1MB is much too big. We're looking right now at 3 lock resources per
inode, so total number of elements depends upon your file system I suppose.
Hopefully I'll have some time today to run some basic tests which may show
us what size strikes a better balance of performance versus memory overhead.
--Mark
--
Mark Fasheh
Senior Software Developer, Oracle
mark.fasheh@oracle.com
next prev parent reply other threads:[~2006-03-05 19:22 UTC|newest]
Thread overview: 57+ messages / expand[flat|nested] mbox.gz Atom feed top
2006-03-03 22:27 [Ocfs2-devel] Ocfs2 performance bugs of doom Daniel Phillips
2006-03-03 22:27 ` Daniel Phillips
2006-03-04 0:53 ` [Ocfs2-devel] " Mark Fasheh
2006-03-04 0:53 ` Mark Fasheh
2006-03-04 3:42 ` [Ocfs2-devel] " Daniel Phillips
2006-03-04 3:42 ` Daniel Phillips
2006-03-04 7:36 ` Andrew Morton
2006-03-04 7:37 ` [Ocfs2-devel] " Andrew Morton
2006-03-05 19:22 ` Mark Fasheh [this message]
2006-03-05 19:22 ` Mark Fasheh
2006-03-06 1:28 ` [Ocfs2-devel] " Daniel Phillips
2006-03-06 1:28 ` Daniel Phillips
2006-03-06 2:58 ` [Ocfs2-devel] " Mark Fasheh
2006-03-06 2:58 ` Mark Fasheh
2006-03-06 4:59 ` Daniel Phillips
2006-03-06 4:59 ` Daniel Phillips
2006-03-06 19:51 ` Mark Fasheh
2006-03-06 19:51 ` Mark Fasheh
2006-03-07 3:34 ` Andi Kleen
2006-03-07 3:34 ` Andi Kleen
2006-03-07 4:58 ` [Ocfs2-devel] " Mark Fasheh
2006-03-07 4:58 ` Mark Fasheh
2006-03-07 6:56 ` [Ocfs2-devel] " Daniel Phillips
2006-03-07 6:56 ` Daniel Phillips
2006-03-09 6:26 ` Daniel Phillips
2006-03-09 6:26 ` Daniel Phillips
2006-03-09 7:26 ` Nick Piggin
2006-03-09 7:26 ` Nick Piggin
2006-03-09 7:43 ` Nick Piggin
2006-03-09 7:43 ` Nick Piggin
2006-03-09 4:19 ` Andi Kleen
2006-03-09 4:19 ` Andi Kleen
2006-03-09 12:30 ` Nick Piggin
2006-03-09 12:30 ` Nick Piggin
2006-03-10 5:14 ` Nick Piggin
2006-03-10 5:14 ` Nick Piggin
2006-03-10 0:21 ` [Ocfs2-devel] Ocfs2 performance Mark Fasheh
2006-03-10 0:21 ` Mark Fasheh
2006-03-10 1:14 ` Bernd Eckenfels
2006-03-10 7:10 ` Joel Becker
2006-03-10 7:10 ` Joel Becker
2006-03-11 1:09 ` Mark Fasheh
2006-03-11 1:09 ` Mark Fasheh
2006-03-11 1:57 ` Bernd Eckenfels
2006-03-11 1:57 ` Bernd Eckenfels
2006-03-10 11:17 ` Daniel Phillips
2006-03-10 11:17 ` Daniel Phillips
2006-03-10 18:23 ` Zach Brown
2006-03-10 18:23 ` Zach Brown
2006-03-10 21:13 ` Daniel Phillips
2006-03-10 21:13 ` Daniel Phillips
2006-03-10 21:13 ` Daniel Phillips
2006-03-10 21:13 ` Daniel Phillips
2006-03-10 2:33 ` [Ocfs2-devel] Ocfs2 performance bugs of doom J. Bruce Fields
2006-03-10 2:33 ` J. Bruce Fields
2006-03-10 10:27 ` Daniel Phillips
2006-03-10 10:27 ` Daniel Phillips
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20060305192247.GH31587@ca-server1.us.oracle.com \
--to=mark.fasheh@oracle.com \
--cc=akpm@osdl.org \
--cc=linux-kernel@vger.kernel.org \
--cc=ocfs2-devel@oss.oracle.com \
--cc=phillips@google.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.