select something from othertable; select * from table where table_id in (?, ?, ?, ?, ?, ?, ?, ...)
This is what MySQL's CEO Martin said in an interview on Slashdot. If we can manage two queries as above through, say, a PHP application, with each executing in 0.004 seconds, then an optimized subquery needs to be beat the 0.008 mark to be a viable alternative. This works for most people who're not looking for elegance of database design. I am assuming MYISAM engine with indexes on the WHERE columns. Granted, this is neither future nor fool proof, but in high traffic environments, it is all worth the drop in elegance. The problem is when the number of rows exceeds 30 million, MySQL performance degrades substantially. For most people, this is not an issue. PG is solid with huge databases, but in my experience, even the most optimized subselect on PG will not return a value in 0.008 seconds on 10 million rows -- I'd appreciate other experiences. Shanx
