From: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
---|---|
To: | David G Johnston <david(dot)g(dot)johnston(at)gmail(dot)com> |
Cc: | pgsql-hackers(at)postgresql(dot)org |
Subject: | Re: Re: [GENERAL] Performance issue with libpq prepared queries on 9.3 and 9.4 |
Date: | 2014-11-14 00:47:24 |
Message-ID: | 27290.1415926044@sss.pgh.pa.us |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-general pgsql-hackers |
David G Johnston <david(dot)g(dot)johnston(at)gmail(dot)com> writes:
> Tom Lane-2 wrote
>> In the meantime, I assume that your real data contains a small percentage
>> of values other than these two? If so, maybe cranking up the statistics
>> target would help. If the planner knows that there are more than two
>> values in the column, I think it would be less optimistic about assuming
>> that the comparison value is one of the big two.
> Is there any value (or can value be added) in creating a partial index of
> the form:
> archetype IN ('banner','some other rare value')
> such that the planner will see that such a value is possible but infrequent
> and will, in the presence of a plan using a value contained in the partial
> index, refuse to use a generic plan knowing that it will be unable to use
> the very specific index that the user created?
The existence of such an index wouldn't alter the planner's statistics.
In theory we could make it do so, but I seriously doubt the cost-benefit
ratio is attractive, either as to implementation effort or the added
planning cost.
regards, tom lane
From | Date | Subject | |
---|---|---|---|
Next Message | David Johnston | 2014-11-14 01:08:35 | Re: Re: [GENERAL] Performance issue with libpq prepared queries on 9.3 and 9.4 |
Previous Message | David G Johnston | 2014-11-14 00:43:45 | Re: [GENERAL] Performance issue with libpq prepared queries on 9.3 and 9.4 |
From | Date | Subject | |
---|---|---|---|
Next Message | Tom Lane | 2014-11-14 00:53:56 | Re: EXPLAIN ANALYZE output weird for Top-N Sort |
Previous Message | Simon Riggs | 2014-11-14 00:46:30 | EXPLAIN ANALYZE output weird for Top-N Sort |