Thread: increase performancr with "noatime"?

increase performancr with "noatime"?

From
JM
Date:
would it cause problem in postgres DB if /var/lib/psql partition is mounted
with "noatime"?

TIA
JM

Re: increase performancr with "noatime"?

From
Andrew Sullivan
Date:
On Tue, Sep 09, 2003 at 04:12:48PM +0800, JM wrote:
> would it cause problem in postgres DB if /var/lib/psql partition is mounted
> with "noatime"?

No; in fact, that's been suggested by many people.  I don't know
whether anyone's done any tests to prove that it helps.

A

--
----
Andrew Sullivan                         204-4141 Yonge Street
Liberty RMS                           Toronto, Ontario Canada
<andrew@libertyrms.info>                              M2P 2A8
                                         +1 416 646 3304 x110


Re: increase performancr with "noatime"?

From
Bill Moran
Date:
Andrew Sullivan wrote:
> On Tue, Sep 09, 2003 at 04:12:48PM +0800, JM wrote:
>
>>would it cause problem in postgres DB if /var/lib/psql partition is mounted
>>with "noatime"?
>
>
> No; in fact, that's been suggested by many people.  I don't know
> whether anyone's done any tests to prove that it helps.

http://www.potentialtech.com/wmoran/postgresql.php#results

You can see, from my _limited_ testing, that it doesn't seem to help enough
to be worth worrying about.  In this test, it actually seems to hurt
performance.  Read the whole page, though.  These tests are heavy on the
writing, it's quite possible that it could improve things if your database
is a heavy read scenerio.

--
Bill Moran
Potential Technologies
http://www.potentialtech.com


Re: increase performancr with "noatime"?

From
Vivek Khera
Date:
>>>>> "AS" == Andrew Sullivan <andrew@libertyrms.info> writes:

AS> On Tue, Sep 09, 2003 at 04:12:48PM +0800, JM wrote:
>> would it cause problem in postgres DB if /var/lib/psql partition is mounted
>> with "noatime"?

AS> No; in fact, that's been suggested by many people.  I don't know
AS> whether anyone's done any tests to prove that it helps.

I honestly can't expect it to be much of an improvement since the
number of files involved compared with the size of the files is
minimal.  However, if you're opening/closing the files often it might
cause you problems.  I think in the normal case where it does matter
you have pooled connections so the open/close happens rarely.

Of course, if I had a good synthetic workload to pound on my DB, I'd
run a test... Sean?

--
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
Vivek Khera, Ph.D.                Khera Communications, Inc.
Internet: khera@kciLink.com       Rockville, MD       +1-240-453-8497
AIM: vivekkhera Y!: vivek_khera   http://www.khera.org/~vivek/