Memcaching Plan of Attack

David Brownlee abs at
Sat Apr 12 10:31:03 BST 2008

On Fri Apr 11, 2008, Jason Tang wrote:
> On Fri Apr 11, 2008 at 12:49:11PM +0100, Simon Wilcox wrote:
> > Jason Tang wrote:
> > >I'm trying to work out if there is anything in the app I dont mind being
> > >out of date. I'm struggling as its a webapp that people use for the
> > >backoffice processes and so everything kind of needs to be up-to-date.
> > 
> > Then you probably don't want a cache. What's the problem you're trying 
> > to solve ?
> Its abit of a juggling game really. Trying to move away raw sql dotted
> all over the repository to something abit more organised - DBIC. In
> converting to DBIC its taken a notable hit on response time. 0.25s -
> 2.5s and SQL DBIC generated is pretty close to the original. I'm
> investigating other avenues (memcached) as well as improving my DBIC-fu.
> I've tried memcached with a 2min TTL and it _might_ be an okay
> compromise. Maybe reduce it a little more.
> Also the existing schema can be much improved but does mean reworking
> alot of legacy stuff. Meh

 	Out of curiosity did you look at Rose::DB::Object when you
 	were moving to an ORM? Its a bit nippier than DBIC and you
 	can turn object caching on where needed (Of course nothing
 	like that is going to beet memcaching generated output, but
 	every little help :)

 	ob "lies, damned lies & benchmark" link

 			   David Brownlee -- abs at

More information about the mailing list