On Wed, 2007-12-12 at 12:56 -0500, Tom Lane wrote:
> Jeff Trout <threshar@threshar.is-a-geek.com> writes:
> > I've seen this on my PITR restores (thankfully, they were for
> > fetching some old data, not because we expoded). On a 2.4ghz opteron
> > it took 5-50 seconds per wal segment, and there were a LOT of
> > segments (replay took hours and hours). I asked a few folks and was
> > told it is the nature of the beast. Hopefully something in 8.4 can
> > be done.
>
> Before we get all panicked about that, someone should try to measure the
> restore speed on 8.3. It's possible that this patch already
> alleviated the problem:
> http://archives.postgresql.org/pgsql-committers/2007-05/msg00041.php
Twas this feature, amongst others, that I had asked to be put back onto
the release notes, so people knew it had improved. I don't really want
to replicate that discussion, but if we're talking about it here then it
probably should be in the notes.
It roughly doubles performance, but there's still more required in some
cases. My recommendation is always measure the performance first.
--
Simon Riggs
2ndQuadrant http://www.2ndQuadrant.com