From: "Ted Jordan" <ted@actionstep.com>
To: "'Merlin Moncure'" <mmoncure@gmail.com>
Cc: <pgsql-performance@postgresql.org>
Subject: Re: Default Performance between 8.0 and 8.1
Date: Tue, 7 Aug 2007 05:58:19 +1200
MIME-Version: 1.0
Content-Type: text/plain;
	charset="us-ascii"
Content-Transfer-Encoding: 7bit
Thread-Index: AcfXzWXTyXecrfGcTTi/cXedMeZoRwAhS0cA
In-Reply-To: <b42b73150708051856g74f886f8h25e42c4b46ec0158@mail.gmail.com>
Message-Id: <20070806175848.D5C8686C80@mail.phase.net.nz>

Thanks Merlin.  I was running both via ssh so it effectively these were
local results.  I will try to isolate the issue to a single statement.  But
I think you have answered the larger question i.e. there is no well known
situation where this happens so I should expect to see roughly the same
performance between 8.0 and 8.1 for data loads and will therefore focus on
the specifics of our environment.

Regards,
-Ted.

-----Original Message-----
From: pgsql-performance-owner@postgresql.org
[mailto:pgsql-performance-owner@postgresql.org] On Behalf Of Merlin Moncure
Sent: 6 August 2007 1:57 p.m.
To: Ted Jordan
Cc: pgsql-performance@postgresql.org
Subject: Re: [PERFORM] Default Performance between 8.0 and 8.1

On 8/5/07, Ted Jordan <ted@actionstep.com> wrote:
> We currently have some databases running under 8.0 and 8.1.  I have
noticed
> a significant difference in the time taken to restore dumps between the
two
> versions, with 8.0 being much faster.  For example a dump of a few MB will
> take just seconds to load under 8.0 but the same dump will take several
> minutes to load under 8.1.  Currently we're just using the default
parameter
> settings when installed via apt on Ubuntu Linux.

a few seconds to several minutes is not the kind of variance that is
explainable by architecture changes between the versions.  something
is amiss.  'maintenance_work_mem' can make a big difference with data
loads due to faster index builds but even that would not account for
the difference.

are you running locally or remotely?  if running remotely, you might
be seeing some networking/tcp problems.

can you reduce your problem down to a single copy statement for analysis?

merlin

---------------------------(end of broadcast)---------------------------
TIP 3: Have you checked our extensive FAQ?

               http://www.postgresql.org/docs/faq