Re: autovacuum_vacuum_cost_delay - Mailing list pgsql-general

From Muhammad Ikram
Subject Re: autovacuum_vacuum_cost_delay
Date
Msg-id CAGeimVqLkSO-PQREPwCgfASz5ENXb9dzLcXNfypEx4J-SPO-DA@mail.gmail.com
Whole thread Raw
In response to autovacuum_vacuum_cost_delay  ("Shenavai, Manuel" <manuel.shenavai@sap.com>)
List pgsql-general
Hi Shenvail,

Based on what I could understand,

Lower value will cause more frequent run of Autovacuum. Will have less delay between processing data chunks but will put more load on the I/O system. Autovacuum tasks will complete faster. Cleanup of dead tuples and blots will be quicker thus increasing performance in the cases where tables are heavily updated/deleted.

Higher values will result in longer delays between processing data chunks thus slowing down the completion of autovacuum tasks. Autovacuum will have less I/O load for its tasks. Database performance will be degraded if operations are more update/delete intensive. Dead tuples and bloat will take longer to be cleaned up. 
It can be beneficial  for performance of other database operations, especially if your system is I/O-bound. Can also be helpful when system resources are less

Regards,
Muhammad Ikram
Bitnine Global.


On Tue, Jul 9, 2024 at 4:38 PM Shenavai, Manuel <manuel.shenavai@sap.com> wrote:

Hi everyone,

 

We are doing some tests with different autovacuum settings.

 

Looking at autovacuum_vacuum_cost_delay:

https://www.postgresql.org/docs/current/runtime-config-autovacuum.html#GUC-AUTOVACUUM-VACUUM-COST-DELAY

 

Can someone help to understand what a high or low value of this setting really means? Would it be OK to set this to 0? If not, why not?

 

Thanks in advance &
Best regards,

Manuel



--
Muhammad Ikram

pgsql-general by date:

Previous
From: "Shenavai, Manuel"
Date:
Subject: autovacuum_vacuum_cost_delay
Next
From: Christoph Moench-Tegeder
Date:
Subject: Re: autovacuum_vacuum_cost_delay