FAQ

Has someone installed a patch to make GEQO work only with > 6 tables,
and changed the SET GEQO to accept a number?

Are we going to do that before 6.1 is released?

Just curious where we are on this.
I'd love to see this in 6.1. I have a series of test queries with 6 tables
differing only in the order they are joined in the where clause.

For 1.0*/6.0, they would run in about 17 and 24 seconds, depending on the
join order used. For 6.1 w/o GEQO, the times are slighly better, about
16 and 22 seconds, and using the r-plans didn't help any more. With GEQO
though and the parameters from the sample file, they all run in 16 and with
a little parameter tweaking, in 12 to 13 seconds!

The GEQO plan I get is _very_ different in its sequence and types of scans,
but it's the same regardless of the join order.

The ability to set this without having to recompile/install would be very,
very beneficial.


darrenk@insightdist.com

------------------------------

Search Discussions

Discussion Posts

Previous

Follow ups

Related Discussions

Discussion Navigation
viewthread | post
posts ‹ prev | 2 of 4 | next ›
Discussion Overview
grouppgsql-hackers @
categoriespostgresql
postedMay 30, '97 at 7:54p
activeMay 31, '97 at 3:29p
posts4
users3
websitepostgresql.org...
irc#postgresql

People

Translate

site design / logo © 2021 Grokbase