From: Richard Huxton <dev@archonet.com>
To: mark@mark.mielke.cc
Cc: Mark Lewis <mark.lewis@mir3.com>
Cc: James Mansion <james@mansionfamily.plus.com>
Cc: Magnus Hagander <magnus@hagander.net>
Cc: Alexander Staubo <alex@purefiction.net>
Cc: Andy <frum@ar-sd.net>
Cc: pgsql-performance@postgresql.org
Subject: Re: LIKE search and performance
Date: Fri, 25 May 2007 16:35:22 +0100
Message-ID: <4657023A.20209@archonet.com> (raw)
In-Reply-To: <20070525141630.GA12162@mark.mielke.cc>
References: <009501c79d4c$445428f0$0b00a8c0@mpsro.dom>
<88daf38c0705230905u1a981102k691b47bd36011a80@mail.gmail.com>
<4655DE75.4080506@mansionfamily.plus.com>
<4655E64B.5050503@hagander.net>
<4655FB8A.1040707@mansionfamily.plus.com>
<1180040560.31471.262.camel@archimedes>
<20070524215450.GA3858@mark.mielke.cc>
<46569AA5.10205@archonet.com>
<20070525141630.GA12162@mark.mielke.cc>
mark@mark.mielke.cc wrote:
> On Fri, May 25, 2007 at 09:13:25AM +0100, Richard Huxton wrote:
>> mark@mark.mielke.cc wrote:
>>>> And since it's basically impossible to know the selectivity of this kind
>>>> of where condition, I doubt the planner would ever realistically want to
>>>> choose that plan anyway because of its poor worst-case behavior.
>>> What is a real life example where an intelligent and researched
>>> database application would issue a like or ilike query as their
>>> primary condition in a situation where they expected very high
>>> selectivity?
>>> Avoiding a poor worst-case behaviour for a worst-case behaviour that
>>> won't happen doesn't seem practical.
>> But if you are also filtering on e.g. date, and that has an index with
>> good selectivity, you're never going to use the text index anyway are
>> you? If you've only got a dozen rows to check against, might as well
>> just read them in.
>> The only time it's worth considering the behaviour at all is *if* the
>> worst-case is possible.
>
> I notice you did not provide a real life example as requested. :-)
OK - any application that allows user-built queries: <choose column:
foo> <choose filter: contains> <choose target: "bar">
Want another? Any application that has a "search by name" box - users
can (and do) put one letter in and hit enter.
Unfortunately you don't always have control over the selectivity of
queries issued.
> This seems like an ivory tower restriction. Not allowing best performance
> in a common situation vs not allowing worst performance in a not-so-common
> situation.
What best performance plan are you thinking of? I'm assuming we're
talking about trailing-wildcard matches here, rather than "contains"
style matches.
--
Richard Huxton
Archonet Ltd
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Reply to all the recipients using the --to and --cc options:
reply via email
To: pgsql-performance@postgresql.org
Cc: dev@archonet.com, mark@mark.mielke.cc, mark.lewis@mir3.com, james@mansionfamily.plus.com, magnus@hagander.net, alex@purefiction.net, frum@ar-sd.net
Subject: Re: LIKE search and performance
In-Reply-To: <4657023A.20209@archonet.com>
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
This inbox is served by DDX for PostgreSQL; see mirroring instructions
for how to clone and mirror all data and code used for this inbox