Re: multi-column index

Tom Lane <tgl@sss.pgh.pa.us>

From: Tom Lane <tgl@sss.pgh.pa.us>
To: David Brown <time@bigpond.net.au>
Cc: Josh Berkus <pgsql-performance@postgresql.org>
Date: 2005-03-17T03:19:13Z
Lists: pgsql-performance
David Brown <time@bigpond.net.au> writes:
> Actually, I'm surprised the planner came up with such a low cost for the 
> single column index, unless ... perhaps correlation statistics aren't 
> used when determining costs for multi-column indexes?

The correlation calculation for multi-column indexes is pretty whacked
out pre-8.0.  I don't think it's that great in 8.0 either --- we really
need to make ANALYZE calculate the correlation explicitly for each
index, probably, rather than trying to use per-column correlations.

			regards, tom lane