I have used Perl DBI and sometimes SQL::Abstract for data-layer. I'd like start to use DBIx::Class or Rose::DB (not decided yet), but I'm not sure about the tunning facilities of this ORM's.
Sometimes, we have complex (very complex) querys writed with no standar SQL (i.e. Oracle rules for using concrete indexes, or PL/SQL calls, etc).
DBI works well for us, perhaps ORM fails in this concrete scenario.
Re:Good note!
grink on 2008-05-11T21:43:06
I haven't encountered any performance problems with DBIx::Class
If I did have a time-critical piece of code and DBIx::Class wasn't up to snuff I'd:
1. Use cursoring (via ->next)
2. Fallback to using $schema->storage->dbh
3. Use HashRefInflator: http://search.cpan.org/user/ash/DBIx-Class-0.08010/lib/DBIx/Class/Manual/Cookbook.po d#Skip_object_creation_for_faster_results
It depends on what you're trying to doRe:Good note!
jjn1056 on 2008-05-13T01:52:03
You can also cache your query results with: http://search.cpan.org/user/mstrout/DBIx-Class-Cursor-Cached-1.0.1/
these works very well if your have a lot of queries that don't often change. You can heavily cache those results and turn it off for your more dynamic queries.
Additionally, there is support for replicated databases and database partitioning. As on the dbix-class IRC channel.
For me, absolute query speed in an abstract sense is not as important as the ability to scale well. That said, I have nothing against other Perl ORMS, just using DBIx-Class very successful on a large social networking site.
Let me know if I can help you with dbic (this is what we call DBIx-Class when we are tired, lazy, etc.)
