From: | Atri Sharma <atri(dot)jiit(at)gmail(dot)com> |
---|---|
To: | David Johnston <polobo(at)yahoo(dot)com> |
Cc: | Pg Hackers <pgsql-hackers(at)postgresql(dot)org> |
Subject: | Re: Planner hints in Postgresql |
Date: | 2014-03-17 17:44:39 |
Message-ID: | CAOeZVieF6BfRB6XYMDfGB+ofqgcZgWdrLitmAfZF4VJBgFp0bQ@mail.gmail.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
The larger question to answer first is whether we want to implement
> something that is deterministic...
>
> How about just dropping the whole concept of "hinting" and provide a way
> for
> someone to say "use this plan, or die trying." Maybe require it be used in
> conjunction with named PREPAREd statements:
>
You mean taking away the entire concept of query planning and cost
estimation? Thats like replacing the optimizer with DBA decision and I am
not at all comfortable with that idea. That are only my thoughts though.
>
> PREPARE s1 (USING /path/to/plan_def_on_server_or_something_similar) AS
> SELECT ...;
>
> Aside from whole-plan specification I can definitely see where join/where
> specification could be useful if it can overcome the current limitation of
> not being able to calculate inter-table estimations.
>
Prepare plans use a generic plan for the execution. Replacing it with a
totally user defined plan does not seem to be clean.
The crux is that IMHO planner hints are a bad way of trying to circumvent
the need for cross-column statistics. We should do cross-column statistics
done and ignore planner hints completely.
Regards,
Atri
--
Regards,
Atri
*l'apprenant*
From | Date | Subject | |
---|---|---|---|
Next Message | Atri Sharma | 2014-03-17 17:45:55 | Re: Planner hints in Postgresql |
Previous Message | Stephen Frost | 2014-03-17 17:43:38 | Re: Planner hints in Postgresql |