By using this site, you agree to our updated Privacy Policy and our Terms of Use. Manage your Cookies Settings.
438,176 Members | 984 Online
Bytes IT Community
+ Ask a Question
Need help? Post your question and get tips & solutions from a community of 438,176 IT Pros & Developers. It's quick & easy.

Reallife szenario for GEQO

P: n/a
I am currently searching some resources, why and where the PostgreSQL
GEQO-Engine improves large join queries. The theoretical background is
clear, but i need some real life szenarios, where the GEQO-Engine beats
other solutions. Have anyone made such a comparison, or have experiences
which szenarios, where the GEQO has improved the query plan generation
in contrast to "traditional" solutions? This will be very helpful (and
interesting) ....

Thanks in advance,

Bernd
---------------------------(end of broadcast)---------------------------
TIP 1: subscribe and unsubscribe commands go to ma*******@postgresql.org

Nov 22 '05 #1
Share this Question
Share on Google+
4 Replies


P: n/a
On Tuesday 13 January 2004 09:50, Bernd Helmle wrote:
Have anyone made such a comparison, or have experiences
which szenarios, where the GEQO has improved the query plan generation
in contrast to "traditional" solutions? This will be very helpful (and
interesting) ....


I don't think it's that it provides a better plan, just that it comes up with
a quicker solution. Searching all possible paths is not practical for a large
number of tables, so you need another approach.

Google for "travelling salesman problem" for discussion of the sort of
thinking behind it.

--
Richard Huxton
Archonet Ltd

---------------------------(end of broadcast)---------------------------
TIP 4: Don't 'kill -9' the postmaster

Nov 22 '05 #2

P: n/a
Richard Huxton wrote:
On Tuesday 13 January 2004 09:50, Bernd Helmle wrote:

I don't think it's that it provides a better plan, just that it comes up with
a quicker solution. Searching all possible paths is not practical for a large
number of tables, so you need another approach.

Aggreed. That was my opinion, too. Sorry for my unclear statement.
Google for "travelling salesman problem" for discussion of the sort of
thinking behind it.


That was the first thing i've done. I understand the theoretical
background, but i need a practical scenario, to show:

QUERY A with GEQO
QUERY A without GEQO

And then compare the specific query plan generation efficency.

In this case I have to think about a practical database scenario, to
execute such queries. If i have enough tables i can join against, i
should see an improvement, in theory.

But what means "enough tables"?

So i wonder, if anyone had done such a comparison already.

Thanks for your reply,

Bernd
---------------------------(end of broadcast)---------------------------
TIP 8: explain analyze is your friend

Nov 22 '05 #3

P: n/a
Bernd Helmle <ma******@oopsware.de> writes:
But what means "enough tables"?


GEQO_THRESHOLD or more. You can set that anywhere you like (I think the
default is 11 or 12, which is about where the standard exhaustive-search
planner becomes painfully slow).

regards, tom lane

---------------------------(end of broadcast)---------------------------
TIP 1: subscribe and unsubscribe commands go to ma*******@postgresql.org

Nov 22 '05 #4

P: n/a
Tom Lane wrote:
Bernd Helmle <ma******@oopsware.de> writes:
But what means "enough tables"?

GEQO_THRESHOLD or more. You can set that anywhere you like (I think the
default is 11 or 12, which is about where the standard exhaustive-search
planner becomes painfully slow).

regards, tom lane


Thank you Tom, that's helpful.

Bernd

---------------------------(end of broadcast)---------------------------
TIP 5: Have you checked our extensive FAQ?

http://www.postgresql.org/docs/faqs/FAQ.html

Nov 22 '05 #5

This discussion thread is closed

Replies have been disabled for this discussion.