Re: [HACKERS] patch: optimize information_schema.constraint_column_usage

The patch applies cleanly. No regression test fails.

On Thu, Feb 2, 2017 at 6:57 PM, Alexey Bashtanov <bashtanov@imap.cc> wrote:
> Hello hackers,
>
> The view information_schema.constraint_column_usage becomes slow when the
> number of columns and constraints raise to substantial values.
> This is because of a join condition that allows only join filter to enforce.
> The patch is to optimize it.

Right. Because there are three relations involved in the condition,
it's evaluation is delayed till all three of them are joined, thus
increasing the sizes of minion joins. With this change, the size of
minion joins reduces. Since this is an all INNER join, splitting the
condition should be fine.

> See many_constraints.sql file attached for a performance test: create 3000
> tables with 10 columns and a PK each and select * from the view.
> The last statement works for 22 seconds on master branch, 34 milliseconds
> optimized on my laptop.

Yes, that's a lot of improvement. I can reproduce similar results on my laptop.

-- 
Best Wishes,
Ashutosh Bapat
EnterpriseDB Corporation
The Postgres Database Company