Re: performance degredation after upgrade from 9.6 to 12

John Felix <[email protected]> Sun, 24 Nov 2019 21:00:48 +0100
Newsgroups gmane.comp.db.postgresql.performance
Message-ID <[email protected]>
Op 24-11-2019 om 19:05 schreef Mariel Cherkassky:
> Hey Jeff,
> This example was only used to show that pg96  had better perfomance 
> than pg12 in a very simple case.
>  In all the tests that I run most of the queries took less time on 
> 9.6`s version.  I dont know why, but as you can see after 
> disabling the parameter the simple test that I did showed different 
> results.  I intend to test this theory tomorrow. I'm going to disable 
> the parameter and run the same application flows that I have on both 
> machines (9.6 vs 12 with zero value for the param).
>
> I didnt send this mail after doing just one simple test, I have more 
> than 100 queries that work better on 9.6 . If u have any explanation I 
> will be happy to hear.
> I'll update tomorrow once I'll have the results..
>
I've had the same experience with parallel query. By default parallel 
query is disabled in 9.6. When we upgraded from 9.6 to 10 is was 
significant slower till I disabled parallel query.

We have a very small database (40 tables and all together max 4GB) and 
we have no long running queries (the largest queries run max 2 to 3 
seconds). That's when parallel query gives a performance degrade in my 
opinion.

Best regards,

John Felix