Horton

Re: Indexes on expressions with multiple columns and operators

Andrei Lepikhov <lepihov@gmail.com>

View thread
From: Andrei Lepikhov <lepihov@gmail.com>
To: Frédéric Yhuel <frederic.yhuel@dalibo.com>, Tom Lane <tgl@sss.pgh.pa.us>
Cc: "pgsql-performance@lists.postgresql.org" <pgsql-performance@lists.postgresql.org>, Jehan-Guillaume de Rorthais <jgdr@dalibo.com>, Christophe Courtois <christophe.courtois@dalibo.com>, Laurenz Albe <laurenz.albe@cybertec.at>
Date: 2025-09-22T21:15:52Z
Lists: pgsql-performance 
On 22/9/2025 18:09, Frédéric Yhuel wrote:
> On 9/22/25 15:57, Andrei Lepikhov wrote:
> I wonder if we could devise another kind of extended statistic that 
> would provide these "partitioned statistics" without actually partitioning.I'm not sure I fully understand your case, but SQL Server demonstrates 
an interesting approach: they have a WHERE clause attached to 
statistics. So, having implemented this, you may separate the whole 
range of values inside the table into 'partitions' by such a WHERE 
condition.
It may solve at least one issue with the 'dependencies' statistics: a 
single number describing the dependency between any two values in the 
columns often leads to incorrect estimations, as I see.

-- 
regards, Andrei Lepikhov