Re: Large number of partitions of a table

From: Victor Sudakov <vas(at)sibptus(dot)ru>
To: pgsql-admin(at)lists(dot)postgresql(dot)org
Cc: Ron <ronljohnsonjr(at)gmail(dot)com>
Subject: Re: Large number of partitions of a table
Date: 2022-01-19 10:27:33
Message-ID: YefnlQRGBnWj4YbF@admin.sibptus.ru
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-admin

Ron wrote:

[dd]

> >
> > What about the system catalogs however? Will the extra 10000
> > tables and 500000 indexes negatively impact the performance of the
> > system catalogs? Are there any caveats you could think of?
>
> EXPLAIN plans are going to be hilariously gigantic, which means that query
> planning would take a loooong time, 

No, I'm not observing this for my test queries. With
enable_partition_pruning=on (which is the default) all the queries
I've tested look neat and short.

> And the query planner (in v12, at
> least) can generate some pretty bad plans in partitioned tables; I bet there
> are edge cases in the QP code that don't work well with 10000 partitions and
> 50000 indices.

Actually "5.11.6. Best Practices for Declarative Partitioning" says
"The query planner is generally able to handle partition hierarchies
with up to a few thousand partitions fairly well, provided that
typical queries allow the query planner to prune all but a small
number of partitions."

I admit 10000 partitions is slightly more than "a few thousand". "A
few" sounds to me like 3-5.

--
Victor Sudakov VAS4-RIPE
http://vas.tomsk.ru/
2:5005/49(at)fidonet

In response to

Browse pgsql-admin by date

  From Date Subject
Next Message Victor Sudakov 2022-01-19 10:30:25 Re: Large number of partitions of a table
Previous Message Ketan Popat 2022-01-19 04:33:46 Re: Large number of partitions of a table