On Sat, Jul 21, 2001 at 12:04:47PM +1000, Philip Warner wrote:
> At 00:49 21/07/01 +1000, Martijn van Oosterhout wrote:
> >- What it does:
> >Adds a new EXPLAIN ANALYZE statement which displays the plan tree, just like
> >EXPLAIN, but also executes the query and collects statistics on what
> >actually happened. The current statistics collected are:
> >
> >Time till first tuple is returned
> >Time till all tuples are returned
> >Total number of tuples returned
> >Number of times Rescan is called
> >
>
> This is a great idea; it's a good first step in getting some stats on how
> accurate the optimizer settings are, as well as how good the assumptions
> are. But you probably need to collect information about IOs & buffer hits
> as well. Perhaps you should look at Jan's stats collection patches and see
> if you can piggyback on them somehow?
Well, the above statistics are done because they can be calculated on each
node. The buffer hits and IOs are harder because they only apply to certain
nodes.
As for Jan's stats sollection patches, where are they? Google finds nothing.
--
Martijn van Oosterhout <kleptog@svana.org>
http://svana.org/kleptog/
> It would be nice if someone came up with a certification system that
> actually separated those who can barely regurgitate what they crammed over
> the last few weeks from those who command secret ninja networking powers.