On Wed, 2002-02-27 at 18:39, Tom Lane wrote:
>
> I cannot believe that caching results for literally-identical queries
> is a win, except perhaps for the most specialized (read brain dead)
> applications.
We have an application which we developed, and which we added
specifically added query caching of the type being discussed. This was
added within the application, rather than within the database, and
provided huge gains for the client.
One difficulty is in knowing when the results of the exact same query
will result in a different result set (i.e. invalidating your cached
resultsets), but as far as I can see this is just as knowable at a
database level as it was for us in the application code.
If this were available I wouldn't want it applied to _all_ queries,
however - I'd want to be able to enable it against particular tables
and/or queries (po: a 'WITH RESULTCACHE' option on the end of a SELECT)
where I knew there was going to be a dramatic win.
The reason there was such a large win for our application was that when
people go to a job website on Monday morning to look for a better place
to work, there is a strong tendency for them to be searching for much
the same sort of thing (as each other) - especially if the site is
getting in the millions of page views.
The application I talk about here was written in Progress (WebSpeed),
and we have found it much harder to do this sort of thing more recently
in Oracle or PostgreSQL.
This would be nice.
Regards, Andrew.
--
--------------------------------------------------------------------
Andrew @ Catalyst .Net.NZ Ltd, PO Box 11-053, Manners St, Wellington
WEB: http://catalyst.net.nz/ PHYS: Level 2, 150-154 Willis St
DDI: +64(4)916-7201 MOB: +64(21)635-694 OFFICE: +64(4)499-2267 Are you enrolled at
http://schoolreunions.co.nz/yet?
