Subject | Re: [Firebird-devel] Re: [IB-Architect] Re: triggers + plans |
---|---|
Author | Jim Starkey |
Post date | 2002-05-22T14:17:19Z |
At 11:36 AM 5/22/02 +0200, Mirza Hadzic wrote:
two alternative indexes, it has a much easier problem to solve than
other database optimizers. The only tricky thing is join order
optimization, made more complicated by the unreasonable requirement
to complete in a feasible amount of time.
The problem with the optimizer isn't that it's dumb. It's broken.
It should be fixed. When it's fixed, it should do a better job
than you can because it knows more about the internal costs of
executing a query.
It hasn't been fixed because there's a workaround -- plans. And,
most likely, as long as plans exist and are extended, it won't
get fixed.
Jim Starkey
>Helo,Because the Interbase/Firebird optimizer never has go chose between
>
>Is it really possible for any optimizer to make (at least almost) perfect
>plan for every query?
>
two alternative indexes, it has a much easier problem to solve than
other database optimizers. The only tricky thing is join order
optimization, made more complicated by the unreasonable requirement
to complete in a feasible amount of time.
The problem with the optimizer isn't that it's dumb. It's broken.
It should be fixed. When it's fixed, it should do a better job
than you can because it knows more about the internal costs of
executing a query.
It hasn't been fixed because there's a workaround -- plans. And,
most likely, as long as plans exist and are extended, it won't
get fixed.
Jim Starkey