On Mon, Dec 16, 2013 at 12:41:43PM -0500, Robert Haas wrote: > Consider a query such as: > > SELECT * FROM a, b WHERE (a.x = 42 AND b.y = 43) OR (a.x = 44 AND b.z = 45); > > If a and/or b are regular tables, the query planner will cleverly > consider the possibility of using an index on a to filter for rows > with a.x = 42 OR a.x = 44, or of using an index on b to filter for > rows where b.y = 43 OR b.z = 45. But if they are foreign tables, this > optimization isn't considered, because we don't intrinsically know > anything about what indexes are present on the foreign side. However, > this optimization could potentially be quite valuable. In fact, it's > arguably more useful here for regular tables, because even if no index > is present on the foreign side, applying the condition on the remote > side might eliminate enough data transfer overhead to win. The only > situation in which I can really see it losing is if the simplified > qual ends up eliminating too few rows to cover the remote side's > processing costs; I'm not sure how possible that is, or how to know > whether it might be the case. > > To see how this can torpedo performance, run the attached SQL file on > an empty database, and then run these quereis:
Advertising
+1 for fixing this bug :) Cheers, David. -- David Fetter <da...@fetter.org> http://fetter.org/ Phone: +1 415 235 3778 AIM: dfetter666 Yahoo!: dfetter Skype: davidfetter XMPP: david.fet...@gmail.com iCal: webcal://www.tripit.com/feed/ical/people/david74/tripit.ics Remember to vote! Consider donating to Postgres: http://www.postgresql.org/about/donate -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
