From: Tom Lane <tgl@sss.pgh.pa.us>
To: Christian Weyer <christian.weyer@thinktecture.com>
Cc: Josh Berkus <josh@agliodbs.com>
Cc: pgsql-performance@postgresql.org <pgsql-performance@postgresql.org>
Subject: Re: Unexpected (bad) performance when querying indexed JSONB column
Date: Sun, 01 Feb 2015 17:20:57 -0500
Message-ID: <4791.1422829257@sss.pgh.pa.us> (raw)
In-Reply-To: <BC4E7C7F-86CB-441A-ABCE-6E0837AA1B65@thinktecture.com>
References: <7F9BEAD7-C190-4E21-8240-04687FA88435@thinktecture.com>
<WM!3b9b688279bfaf852b09eb38e12d59633671a1b1dacc67e4f7be92553c116d3655cd62355ebe242eb4ae45275d2a5848!@asav-2.01.com>
<54CE954F.7070904@agliodbs.com>
<BC4E7C7F-86CB-441A-ABCE-6E0837AA1B65@thinktecture.com>
List-Unsubscribe: <mailto:majordomo@postgresql.org?body=unsub%20pgsql-performance>
Christian Weyer <christian.weyer@thinktecture.com> writes:
> On 01.02.15 22:06, "Josh Berkus" <josh@agliodbs.com> wrote:
>> Please send us the output of EXPLAIN ( ANALYZE ON, BUFFERS ON ) so that
>> we can see what the query is actually doing, rather than just what the
>> plan was.
> Sure. Here we go:
> "Bitmap Heap Scan on articles (cost=16.25..135.64 rows=33 width=427)
> (actual time=6.425..43.603 rows=18584 loops=1)"
> " Recheck Cond: (data @> ‘{"locked": true}'::jsonb)"
> " Heap Blocks: exact=1496"
> " Buffers: shared hit=1504"
> " -> Bitmap Index Scan on idx_data (cost=0.00..16.24 rows=33 width=0)
> (actual time=6.090..6.090 rows=18584 loops=1)"
> " Index Cond: (data @> ‘{"locked": true}'::jsonb)"
> " Buffers: shared hit=8"
> "Planning time: 0.348 ms"
> "Execution time: 47.788 ms"
So that's showing a runtime of 48 ms, not 900. For retrieving 18584
rows, doesn't sound that bad to me.
(If the planner had had a better rowcount estimate, it'd likely have
not bothered with the index at all but just done a seqscan. This is
a consequence of the lack of any very useful stats for JSONB columns,
which is something we hope to address soon; but it's not done in 9.4
and likely won't be in 9.5 either ...)
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, christian.weyer@thinktecture.com, josh@agliodbs.com
Subject: Re: Unexpected (bad) performance when querying indexed JSONB column
In-Reply-To: <4791.1422829257@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