Subject: Re: Searching for the cause of a bad plan
From: Simon Riggs <simon@2ndquadrant.com>
To: Tom Lane <tgl@sss.pgh.pa.us>
Cc: Csaba Nagy <nagy@ecircle-ag.com>,
	postgres performance list <pgsql-performance@postgresql.org>
In-Reply-To: <29186.1190904025@sss.pgh.pa.us>
References: <1190369024.4661.173.camel@PCD12478>
	<6175.1190390913@sss.pgh.pa.us> <1190396352.4202.109.camel@ebony.site>
	<13644.1190397220@sss.pgh.pa.us> <1190399879.4202.127.camel@ebony.site>
	<27857.1190417434@sss.pgh.pa.us> <1190640429.4181.165.camel@ebony.site>
	<1190818357.5430.29.camel@PCD12478> <21582.1190820137@sss.pgh.pa.us>
	<1190894004.5430.46.camel@PCD12478>  <29186.1190904025@sss.pgh.pa.us>
Content-Type: text/plain
Date: Thu, 27 Sep 2007 16:24:37 +0100
Message-Id: <1190906677.4194.78.camel@ebony.site>
Mime-Version: 1.0
Content-Transfer-Encoding: 7bit

On Thu, 2007-09-27 at 10:40 -0400, Tom Lane wrote:

> And yet there's another trap here: if the parameter you passed in
> chanced to be one of the very common values, a plan that was optimized
> for a small number of matches would perform terribly.

I wonder could we move prepare_threshold onto the server? (from JDBC).

When we prepare a statement, we record the current setting of the
prepare_threshold parameter into the plan for that statement. Valid
settings are -1, 0 or more.
If the setting is -1 then we always re-plan the statement. 
If the setting is 0 then we plan the statement generically.
If the setting is > 0 then we plan the statement according to the
specific parameter value and then decrement the prepare count, so
eventually the executions will hit 0 (above).

That would also simplify JDBC and allow the functionality from other
clients, as well as from within PLs.

We could also force the plan to be -1 when the planning involves
something that would force it to be a one-time plan, e.g. constraint
exclusion. (So we could then get rid of that parameter).

-- 
  Simon Riggs
  2ndQuadrant  http://www.2ndQuadrant.com