pg.ddx.io  pgsql-performance@postgresql.org mailing list archive  
help / color / mirror / Atom feed
Comparing postgresql7.4 CVS head on linux 2.4.20 and 2.6.0-test4
3+ messages / 2 participants
[nested] [flat]

* Comparing postgresql7.4 CVS head on linux 2.4.20 and 2.6.0-test4
@ 2003-08-27 15:32  Shridhar Daithankar <shridhar_daithankar@persistent.co.in>
  0 siblings, 2 replies; 3+ messages in thread

From: Shridhar Daithankar @ 2003-08-27 15:32 UTC (permalink / raw)
  To: pgsql-performance

Hi all,

I did some benchmarking using pgbench and postgresql CVS head, compiled 
yesterday. 

The results are attached. It looks like 2.6.0-test4 does better under load but 
under light load the performance isn't that great. OTOH 2.4.20 suffer major 
degradation compare to 2.6. Looks like linux is also getting heavy at lower 
end. Of course it isn't as bad as solaris as yet..:-)

IIRC in a kernel release note recently, it was commented that IO scheduler is 
still being worked on and does not perform as much for random seeks, which 
exaclty what database needs.

How does these number stack up with other platforms? Anybody with SCSI disks 
out there? I doubt IDE has some role to play with this.

Comments?

 Shridhar

Attachments:

pgbench.7.4CVSHEAD.2.6.0-test4 (text/plain, 2.4K) download Vacumm full done after each run. pg_resetlog run between each server start Scaling factor: 10 Initialization time: real 1m42.282s user 0m1.787s sys 0m0.121s Kernel 2.6.0-test4 Effective cache size 38400 (300MB) checkpoint segments=6 1)Shared buffers 1000 pgbench -c5 -t100 test1 tps = 131.962542 (including connections establishing) tps = 134.418358 (excluding connections establishing) pgbench -c5 -t1000 test1 tps = 79.301493 (including connections establishing) tps = 79.393258 (excluding connections establishing) pgbench -c10 -t100 test1 tps = 64.917044 (including connections establishing) tps = 65.438067 (excluding connections establishing) pgbench -c10 -t1000 test1 tps = 66.135615 (including connections establishing) tps = 66.193039 (excluding connections establishing) 2) Shared buffers 2000 pgbench -c5 -t100 test1 tps = 135.598994 (including connections establishing) tps = 138.296634 (excluding connections establishing) pgbench -c5 -t1000 test1 tps = 81.331678 (including connections establishing) tps = 81.425369 (excluding connections establishing) pgbench -c10 -t100 test1 tps = 116.524476 (including connections establishing) tps = 118.388407 (excluding connections establishing) pgbench -c10 -t1000 test1 tps = 64.659894 (including connections establishing) tps = 64.719829 (excluding connections establishing) 3) Shared buffers 3000 pgbench -c5 -t100 test tps = 132.489569 (including connections establishing) tps = 135.177003 (excluding connections establishing) pgbench -c5 -t1000 test tps = 70.272855 (including connections establishing) tps = 70.343452 (excluding connections establishing) pgbench -c10 -t100 test tps = 121.624524 (including connections establishing) tps = 123.549086 (excluding connections establishing) pgbench -c10 -t1000 test tps = 64.124644 (including connections establishing) tps = 64.183736 (excluding connections establishing) 4) noatime enabled Shared buffers 3000 pgbench -c5 -t100 test tps = 90.850600 (including connections establishing) tps = 92.053686 (excluding connections establishing) pgbench -c5 -t1000 test tps = 92.209724 (including connections establishing) tps = 92.329682 (excluding connections establishing) pgbench -c10 -t100 test tps = 79.264231 (including connections establishing) tps = 80.145448 (excluding connections establishing) pgbench -c10 -t1000 test tps = 70.439082 (including connections establishing) tps = 70.506189 (excluding connections establishing)
pgbench.7.4CVSHEAD.24.20 (text/plain, 2.4K) download Vacumm full done after each run. pg_resetlog run between each server start Scaling factor: 10 Initialization time: real 1m32.551s user 0m1.710s sys 0m0.130s Kernel 2.4.20 Effective cache size 38400 (300MB) checkpoint segments=6 1)Shared buffers 1000 pgbench -c5 -t100 test1 tps = 145.504917 (including connections establishing) tps = 147.991299 (excluding connections establishing) pgbench -c5 -t1000 test1 tps = 70.785341 (including connections establishing) tps = 70.849065 (excluding connections establishing) pgbench -c10 -t100 test1 tps = 147.605267 (including connections establishing) tps = 149.944123 (excluding connections establishing) pgbench -c10 -t1000 test1 tps = 55.646866 (including connections establishing) tps = 55.678399 (excluding connections establishing) 2) Shared buffers 2000 pgbench -c5 -t100 test1 tps = 140.115444 (including connections establishing) tps = 142.441377 (excluding connections establishing) pgbench -c5 -t1000 test1 tps = 77.015876 (including connections establishing) tps = 77.084474 (excluding connections establishing) pgbench -c10 -t100 test1 tps = 128.512768 (including connections establishing) tps = 130.314689 (excluding connections establishing) pgbench -c10 -t1000 test1 tps = 55.003077 (including connections establishing) tps = 55.035001 (excluding connections establishing) 3) Shared buffers 3000 pgbench -c5 -t100 test1 tps = 153.142499 (including connections establishing) tps = 156.094637 (excluding connections establishing) pgbench -c5 -t1000 test1 tps = 81.391878 (including connections establishing) tps = 81.472845 (excluding connections establishing) pgbench -c10 -t100 test1 tps = 120.420876 (including connections establishing) tps = 122.066975 (excluding connections establishing) pgbench -c10 -t1000 test1 tps = 56.766460 (including connections establishing) tps = 56.801117 (excluding connections establishing) 4) Shared buffers 3000, noatime pgbench -c5 -t100 test1 tps = 140.464589 (including connections establishing) tps = 143.152323 (excluding connections establishing) pgbench -c5 -t1000 test1 tps = 94.718161 (including connections establishing) tps = 94.826618 (excluding connections establishing) pgbench -c10 -t100 test1 tps = 131.925129 (including connections establishing) tps = 133.826403 (excluding connections establishing) pgbench -c10 -t1000 test1 tps = 65.374006 (including connections establishing) tps = 65.458626 (excluding connections establishing)
^ permalink raw reply [nested|flat] 3+ messages in thread

* Re: Comparing postgresql7.4 CVS head on linux 2.4.20 and 2.6.0-test4
@ 2003-08-27 23:00  Neil Conway <neilc@samurai.com>
  parent: Shridhar Daithankar <shridhar_daithankar@persistent.co.in>
  1 sibling, 0 replies; 3+ messages in thread

From: Neil Conway @ 2003-08-27 23:00 UTC (permalink / raw)
  To: Shridhar Daithankar <shridhar_daithankar@persistent.co.in>; +Cc: pgsql-performance

On Wed, Aug 27, 2003 at 09:02:25PM +0530, Shridhar Daithankar wrote:
> IIRC in a kernel release note recently, it was commented that IO scheduler is 
> still being worked on and does not perform as much for random seeks, which 
> exaclty what database needs.

Yeah, I've read that as well. It would be interesting to see how 2.6
performs with the traditional (non-anticipatory) scheduler -- I believe
you can switch from one I/O scheduler to another via a sysctl.

> pgbench -c10 -t100 test1
> tps = 64.917044 (including connections establishing)
> tps = 65.438067 (excluding connections establishing)

Interesting that the performance of 2.4.20 for this particular
benchmark is a little less than 3 times faster than 2.6

> 3) Shared buffers 3000
> 
> pgbench -c5 -t100 test
> tps = 132.489569 (including connections establishing)
> tps = 135.177003 (excluding connections establishing)
> 
> pgbench -c5 -t1000 test
> tps = 70.272855 (including connections establishing)
> tps = 70.343452 (excluding connections establishing)
> 
> pgbench -c10 -t100 test
> tps = 121.624524 (including connections establishing)
> tps = 123.549086 (excluding connections establishing)

[...] 

> 4) noatime enabled Shared buffers 3000
> 
> pgbench -c5 -t100 test
> tps = 90.850600 (including connections establishing)
> tps = 92.053686 (excluding connections establishing)
> 
> pgbench -c5 -t1000 test
> tps = 92.209724 (including connections establishing)
> tps = 92.329682 (excluding connections establishing)
> 
> pgbench -c10 -t100 test
> tps = 79.264231 (including connections establishing)
> tps = 80.145448 (excluding connections establishing)

I'm a little skeptical of the consistency of these numbers
(several people have observed in the past that it's difficult
to get pgbench to produce reliable results) -- how is it
possible that using noatime can possibly *reduce* performance
by 50%, in the case of the first and third benchmarks?

-Neil




^ permalink  raw  reply  [nested|flat] 3+ messages in thread

* Re: Comparing postgresql7.4 CVS head on linux 2.4.20 and 2.6.0-test4
@ 2003-08-28 06:53  Shridhar Daithankar <shridhar_daithankar@persistent.co.in>
  parent: Shridhar Daithankar <shridhar_daithankar@persistent.co.in>
  1 sibling, 0 replies; 3+ messages in thread

From: Shridhar Daithankar @ 2003-08-28 06:53 UTC (permalink / raw)
  To: pgsql-performance

On 27 Aug 2003 at 19:00, Neil Conway wrote:

> On Wed, Aug 27, 2003 at 09:02:25PM +0530, Shridhar Daithankar wrote:
> > IIRC in a kernel release note recently, it was commented that IO scheduler is 
> > still being worked on and does not perform as much for random seeks, which 
> > exaclty what database needs.
> 
> Yeah, I've read that as well. It would be interesting to see how 2.6
> performs with the traditional (non-anticipatory) scheduler -- I believe
> you can switch from one I/O scheduler to another via a sysctl.

I will repeat the tests after get that setting. Will google for it..

> 
> > pgbench -c10 -t100 test1
> > tps = 64.917044 (including connections establishing)
> > tps = 65.438067 (excluding connections establishing)
> 
> Interesting that the performance of 2.4.20 for this particular
> benchmark is a little less than 3 times faster than 2.6

Yeah but 2.4 drops like anything..
> > 4) noatime enabled Shared buffers 3000
> > 
> > pgbench -c5 -t100 test
> > tps = 90.850600 (including connections establishing)
> > tps = 92.053686 (excluding connections establishing)
> > 
> > pgbench -c5 -t1000 test
> > tps = 92.209724 (including connections establishing)
> > tps = 92.329682 (excluding connections establishing)
> > 
> > pgbench -c10 -t100 test
> > tps = 79.264231 (including connections establishing)
> > tps = 80.145448 (excluding connections establishing)
> 
> I'm a little skeptical of the consistency of these numbers
> (several people have observed in the past that it's difficult
> to get pgbench to produce reliable results) -- how is it
> possible that using noatime can possibly *reduce* performance
> by 50%, in the case of the first and third benchmarks?

I know. I am puzzled too. Probably I didn't put noatime properly in /etc/fstab. 
Unfortunately I have only one linux partition. So I prefer to boot rather than 
remounting root.

I will redo the bechmarks and post the results..

Bye
 Shridhar

--
Law of Communications:	The inevitable result of improved and enlarged 
communications	between different levels in a hierarchy is a vastly increased	
area of misunderstanding.




^ permalink  raw  reply  [nested|flat] 3+ messages in thread


end of thread, other threads:[~2003-08-28 06:53 UTC | newest]

Thread overview: 3+ messages (download: mbox mbox.gz follow: Atom feed)
-- links below jump to the message on this page --
2003-08-27 15:32 Comparing postgresql7.4 CVS head on linux 2.4.20 and 2.6.0-test4 Shridhar Daithankar <shridhar_daithankar@persistent.co.in>
2003-08-27 23:00 ` Neil Conway <neilc@samurai.com>
2003-08-28 06:53 ` Shridhar Daithankar <shridhar_daithankar@persistent.co.in>

This inbox is served by DDX for PostgreSQL; see mirroring instructions
for how to clone and mirror all data and code used for this inbox