Curmudgeonly

After having spent the last two years working with MySQL I think it’s finally time to say ‘NO’. The problem is that as an RDBMS is just doesn’t perform, don’t get my wrong it can do somethings incredibly quickly. Like select a row based on primary key. But give it any kind of JOINs and conditions and poof, it’s now slower than molasses.

The problem is that the mysql team is clearly not interested in fixing the problems with their query optimizer. Clearly the push to make MySQL more ‘enterprise’ ready is taking precedence over basic performance. The support team’s only response is to blow smoke.

Unfortunately I can’t just write (or steal from a working RDBMS) a new optimizer. Because MySQL AB want to sell commercial licenses I can’t just write an optimizer under GPL and get it into the source tree. So my only option is to constantly rewrite the same old SQL to make it perform. In effect writing the world first external query optimizer.

Enough is enough. I say it’s time to fork MySQL and to create a db that actually meets all the basic requirements before we go adding in stored procs and such. Things like basic ACID compliance and decent query performance.