|Perl Monk, Perl Meditation
Re^3: perl & SQL best practices (joins)by tye (Sage)
|on Apr 27, 2012 at 23:33 UTC
By all means, start out doing table joins inside the DB. I have relatively rarely seen a situation develop where the schema vs. the needed data (and how often and how quickly) ends up showing how you've painted yourself into a bit of a corner and you can work around the problem by doing some joins in the client.
Much of the time, this is also about limitations or quirks of the query optimizer and how much time / expertise you can throw at recrafting the query to convince the optimizer to do the efficient thing.
Frankly, I've more often seen client-side joins done because the people working on the project rather sucked at database performance issues. Or even because the team just believed that "joins are expensive". But when I've actually worked on the performance issue that lead to client-side joins, there was a much more efficient solution that didn't involve client-side joins.
So, I have done client-side joins for performance reasons. But only very rarely and always due to things that had better resolutions in the long run. It can be a powerful work-around.
I've never seen client-side joins done at the outset for good reasons. So, yes, start out with server-side joins, by all means.