Message-ID: <4EAC10D4.1080300@comodo.com>
Date: Sat, 29 Oct 2011 20:12:28 +0530
From: Robins Tharakan <robins.tharakan@comodo.com>
Organization: Comodo
User-Agent: Mozilla/5.0 (X11; U; Linux x86_64; en-US;
	rv:1.9.2.23) Gecko/20110922 Thunderbird/3.1.15
MIME-Version: 1.0
To: pgsql-performance@postgresql.org
Subject: Re: Bad plan by Planner (Already resolved?)
References: <4E9D2ECA0200002500042131@gw.wicourts.gov>
	<4EA6952E.90209@comodo.com> <9507.1319659042@sss.pgh.pa.us>
In-Reply-To: <9507.1319659042@sss.pgh.pa.us>
Content-Type: multipart/signed; protocol="application/pkcs7-signature";
	micalg=sha1; boundary="------------ms010800050704070008010208"

This is a cryptographically signed message in MIME format.

--------------ms010800050704070008010208
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: quoted-printable

Thanks Tom!

Regret the delay in reply, but two of the three guesses were spot-on and =

resolved the doubt. 8.4.9 does take care of this case very well.

On 10/27/2011 01:27 AM, Tom Lane wrote:
> I suspect that you're just fooling yourself here, and the "optimized"
> query is no such thing.
:) I actually meant 'faster' query, but well...


> 1. The tables are horrendously bloated on the first database, so that
> many more pages have to be touched to get the same number of tuples.
> This would likely indicate an improper autovacuum configuration.
I believe you've nailed it pretty accurately. The tables are=20
horrendously bloated and I may need to tune AutoVacuum to be much more=20
aggressive than it is. I did see that HashAggregate makes only a minor=20
difference, but what didn't strike is that the slowness could be bloat.


> 2. You failed to account for caching effects, ie the first example
> is being run "cold" and has to actually read everything from disk,
> whereas the second example has everything it needs already in RAM.
> In that case the speed differential is quite illusory.
On hindsight, this was a miss. Should have warmed the caches before=20
posting. Re-running this query multiple times, brought out the result in =

~100ms.

> BTW, how come is it that "SELECT large_table_b.field_b FROM
> large_table_b WHERE field_a =3D 2673056" produces no duplicate field_b
> values?  Is that just luck?  Is there a unique constraint on the table
> that implies it will happen?
Its just luck. Sometimes the corresponding values genuinely don't exist=20
in the other table, so that's ok.

--
Robins Tharakan


--------------ms010800050704070008010208
Content-Type: application/pkcs7-signature; name="smime.p7s"
Content-Transfer-Encoding: base64
Content-Disposition: attachment; filename="smime.p7s"
Content-Description: S/MIME Cryptographic Signature
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--------------ms010800050704070008010208--