Re: How to speed up min/max(id) in 50M rows table? - Mailing list pgsql-performance

From Alan Hodgson
Subject Re: How to speed up min/max(id) in 50M rows table?
Date
Msg-id 200710121504.40365@hal.medialogik.com
Whole thread Raw
In response to Re: How to speed up min/max(id) in 50M rows table?  (henk de wit <henk53602@hotmail.com>)
List pgsql-performance
On Friday 12 October 2007, henk de wit <henk53602@hotmail.com> wrote:
> >   select payment_id from transactions order by payment_id desc limit 1;
>
> This one is indeed instant! Less than 50ms. In my case I can't use it for
> max though because of the fact that payment_id can be null (which is an
> unfortunate design choice). The other variant however didn't become
> instant. I.e. I tried:
>
> select time from transactions where payment_id = 67 order by time asc
> limit 1;
>
> But this one is still really slow.

If you had a compound index on payment_id,time (especially with a WHERE time
IS NOT NULL conditional) it would likely speed it up.


--
Ghawar is dying


pgsql-performance by date:

Previous
From: Tom Lane
Date:
Subject: Re: How to speed up min/max(id) in 50M rows table?
Next
From: Erik Jones
Date:
Subject: Re: Huge amount of memory consumed during transaction