On Tue, 2010-02-23 at 11:40 -0500, Andrew Sullivan wrote:
> On Tue, Feb 23, 2010 at 03:09:49PM +0700, Stuart Bishop wrote:
> > The slony information is interesting as I don't need to take 60 seconds to get a 60 second delta, which would let me serve these stats from a web application without some daemon in the background constantly refreshing a cache. I'm just exploring at this stage anyway.
> >
> 
> But the statio information is way cheaper and faster to get.  Doing
> the queries you're thinking of will cause you disk use you maybe don't
> want to pay.

And infinitely more useful.

Grovelling through Slony will tell you that you are doing X number of
updates.  Then what?

Are they HOT updates or live updates?  Are they writing a handful of
blocks or lots and lots of blocks?  Are those blocks getting written
once at checkpoint time or repeatedly by the bgwriter?  This is the sort
of information you can get from the Postgres views that you won't get
from Slony.  It's the sort of data you'll need to really pinpoint what
is going on and how to fix it.

-- 
Brad Nicholson  416-673-4106
Database Administrator, Afilias Canada Corp.