hello everybody,

we came across an issue which turned out to be more serious than previously expected.
imagine a system with, say, 1000 partitions (heavily indexed) or so. the time taken by the planner is already fairly heavy in this case.

i tried this one with 5000 unindexed tables (just one col):

test=# \timing
Timing is on.
test=# prepare x(int4) AS select * from t_data order by id desc;
PREPARE
Time: 361.552 ms

you will see similar or higher runtimes in case of 500 partitions and a handful of indexes.

does anybody see a potential way to do a shortcut through the planner?
a prepared query is no solution here as constraint exclusion would be dead in this case (making the entire thing an even bigger drama).

did anybody think of a solution to this problem.
or more precisely: can there be a solution to this problem?

	many thanks,

		hans

--
Cybertec Schönig & Schönig GmbH
Gröhrmühlgasse 26
A-2700 Wiener Neustadt, Austria
Web: http://www.postgresql-support.de

Responses

• Re: plan time of MASSIVE partitioning ... at 2010-09-03 12:04:42 from Stephen Frost
• Re: plan time of MASSIVE partitioning ... at 2010-09-03 13:27:15 from Robert Haas
• Re: plan time of MASSIVE partitioning ... at 2010-09-03 14:40:54 from Tom Lane

pgsql-hackers by date

Next:	From: Boszormenyi Zoltan	Date: 2010-09-03 09:43:32
	Subject: Re: Synchronization levels in SR
Previous:	From: Heikki Linnakangas	Date: 2010-09-03 09:33:43
	Subject: Re: Synchronous replication - patch status inquiry