From: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
---|---|
To: | Stephen Frost <sfrost(at)snowman(dot)net> |
Cc: | Job <Job(at)colliniconsulting(dot)it>, Jaime Soler <jaime(dot)soler(at)gmail(dot)com>, "pgsql-general(at)postgresql(dot)org" <pgsql-general(at)postgresql(dot)org> |
Subject: | Re: R: R: Slow queries on very big (and partitioned) table |
Date: | 2017-02-21 23:11:17 |
Message-ID: | 16979.1487718677@sss.pgh.pa.us |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-general |
Stephen Frost <sfrost(at)snowman(dot)net> writes:
> * Job (Job(at)colliniconsulting(dot)it) wrote:
>> We thought to implement one partition for day.
>> Do you think it should be fine?
> Really depends on what you're doing. If you're running very short
> queries that pull out just a record or a few records, then you're going
> to be unhappy with the planning time required when you have hundreds and
> thousands of partitions, which is why I typically recommend against
> using partitions-by-day unless you're only keeping a few months worth of
> data.
Or to put it more simply: if you have more than O(100) partitions,
you're doing it wrong. There is a cost to subdividing things too finely.
The improved partitioning support that's going into v10 will probably
allow more partitions before it really starts to groan, but it'll still
not be a great idea to create more than the minimum number of partitions
you really need.
regards, tom lane
From | Date | Subject | |
---|---|---|---|
Next Message | Patrick B | 2017-02-21 23:41:34 | Re: bloat indexes - opinion |
Previous Message | David G. Johnston | 2017-02-21 23:05:38 | Re: NOTIFY command impact |