Re: Deleting older versions in unique indexes to avoid page splits - Mailing list pgsql-hackers

From Victor Yegorov
Subject Re: Deleting older versions in unique indexes to avoid page splits
Date
Msg-id CAGnEbogr0jOzrPQYFN16TQy3BYWW0GgwzGE3i_MWkcA3qwNXpw@mail.gmail.com
Whole thread Raw
In response to Re: Deleting older versions in unique indexes to avoid page splits  (Peter Geoghegan <pg@bowt.ie>)
Responses Re: Deleting older versions in unique indexes to avoid page splits  (Peter Geoghegan <pg@bowt.ie>)
List pgsql-hackers
пт, 16 окт. 2020 г. в 21:12, Peter Geoghegan <pg@bowt.ie>:
I ran the script for two hours and 16 clients with the patch, then for
another two hours with master. After that time, all 3 indexes were
exactly the same size with the patch, but had almost doubled in size
on master:

aid_pkey_include_abalance: 784,264 pages (or ~5.983 GiB)
one: 769,545 pages (or ~5.871 GiB)
two: 770,295 pages (or ~5.876 GiB)

(With the patch, all three indexes were 100% pristine -- they remained
precisely 411,289 pages in size by the end, which is ~3.137 GiB.)



The TPS/throughput is about what you'd expect for the two hour run:

18,988.762398 TPS for the patch
11,123.551707 TPS for the master branch.

I really like these results, great work!

I'm also wondering how IO numbers changed due to these improvements, shouldn't be difficult to look into.

Peter, according to cfbot patch no longer compiles.
Can you send and update, please?


--
Victor Yegorov

pgsql-hackers by date:

Previous
From: Peter Geoghegan
Date:
Subject: Re: More aggressive vacuuming of temporary tables
Next
From: Sergei Kornilov
Date:
Subject: Re: allow online change primary_conninfo