From: Tom Lane <tgl@sss.pgh.pa.us>
To: AlexK987 <alex.cue.987@gmail.com>
Cc: pgsql-performance@postgresql.org
Subject: Re: Why is GIN index slowing down my query?
Date: Sun, 01 Feb 2015 22:15:45 -0500
Message-ID: <12830.1422846945@sss.pgh.pa.us> (raw)
In-Reply-To: <1422841046401-5836323.post@n5.nabble.com>
References: <1422829941197-5836319.post@n5.nabble.com>
<6888.1422833653@sss.pgh.pa.us>
<1422841046401-5836323.post@n5.nabble.com>
List-Unsubscribe: <mailto:majordomo@postgresql.org?body=unsub%20pgsql-performance>
AlexK987 <alex.cue.987@gmail.com> writes:
> This is a realistic case: everyone have Python and Java skills, but PostGis
> and Haskell and Closure are rare. If we are looking for a person that has
> all the skills required for a task (array[1, 15]), that is "skills <@
> array[1, 15] " and not the opposite, right?
One of us has this backwards. It might be me, but I don't think so.
Consider a person who has the two desired skills plus skill #42:
regression=# select array[1,15,42] <@ array[1,15];
?column?
----------
f
(1 row)
regression=# select array[1,15,42] @> array[1,15];
?column?
----------
t
(1 row)
> Also can you explain why " entries for "0" and "1" swamp everything else so
> that the planner
> doesn't know that eg "15" is really rare. " I thought that if a value is not
> found in the histogram, than clearly that value is rare, correct? What am I
> missing here?
The problem is *how* rare. The planner will take the lowest frequency
seen among the most common elements as an upper bound for the frequency of
unlisted elements --- but if all you have in the stats array is 0 and 1,
and they both have frequency 1.0, that doesn't tell you anything. And
that's what I see for this example:
regression=# select most_common_elems,most_common_elem_freqs from pg_stats where tablename = 'talent' and attname = 'skills';
most_common_elems | most_common_elem_freqs
-------------------+------------------------
{0,1} | {1,1,1,1,0}
(1 row)
With a less skewed distribution, that rule of thumb would work better :-(
regards, tom lane
--
Sent via pgsql-performance mailing list (pgsql-performance@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-performance
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: tgl@sss.pgh.pa.us, alex.cue.987@gmail.com
Subject: Re: Why is GIN index slowing down my query?
In-Reply-To: <12830.1422846945@sss.pgh.pa.us>
* 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