Re: Further pg_upgrade analysis for many tables

On Sat, Nov 10, 2012 at 05:59:54PM -0500, Bruce Momjian wrote:
> On Sat, Nov 10, 2012 at 02:45:54PM -0800, Jeff Janes wrote:
> > On Sat, Nov 10, 2012 at 9:15 AM, Bruce Momjian <bruce(at)momjian(dot)us> wrote:
> > > On Fri, Nov 9, 2012 at 04:23:40PM -0800, Jeff Janes wrote:
> > >> On Fri, Nov 9, 2012 at 3:06 PM, Bruce Momjian <bruce(at)momjian(dot)us> wrote:
> > >> >
> > >> > Again, using SERIAL?
> > >>
> > >> Yep.
> > >
> > > Odd why yours is so much after.
> >
> > You didn't build git head under --enable-cassert, did you?
>
> Yikes, you got me! I have not done performance testing in so long, I
> had forgotten I changed my defaults. New numbers to follow. Sorry.
>
> > Any chance you can do a oprofile or gprof of head's pg_dump dumping
> > out of head's server? That really should be a lot faster (since
> > commit eeb6f37d89fc60c6449ca12ef9e) than dumping out of 9.2 server.
> > If it is not for you, I don't see how to figure it out without a
> > profile of the slow system.
>
> Yes, coming.

OK, here are my results. Again, apologies for posting non-linear
results based on assert builds:

---------- 9.2 ---------- ------------ 9.3 --------
-- normal -- -- bin-up -- -- normal -- -- bin-up --
dump rest dump rest dump rest dump rest pg_upgrade
1 0.12 0.06 0.12 0.06 0.11 0.07 0.11 0.07 11.11
1000 7.22 2.40 4.74 2.78 2.20 2.43 4.04 2.86 19.60
2000 5.67 5.10 8.82 5.57 4.50 4.97 8.07 5.69 30.55
4000 13.34 11.13 25.16 12.52 8.95 11.24 16.75 12.16 60.70
8000 29.12 25.98 59.60 28.08 16.68 24.02 30.63 27.08 123.05
16000 87.36 53.16 189.38 62.72 31.38 55.37 61.55 62.66 365.71

You can see the non-linear dump at 16k in 9.2, and the almost-linear in
9.3. :-)

pg_upgrade shows non-linear, but that is probably because of the
non-linear behavior of 9.2, and because of the two non-linear loops that
Ants found, that I will address in a separate email.

Thanks for the feedback.

--
Bruce Momjian <bruce(at)momjian(dot)us> http://momjian.us
EnterpriseDB http://enterprisedb.com

+ It's impossible for everything to be true. +