I can see all of the postgresql indices/primary keys are created as btree AND I have also noted that a lot of the tables uses UUID as a primary key - but you are still using UUIDv4 values which due to their truly random nature are spread all over the range of valid values.
Which means that even values that are more or less timebased such as job_sessions, job_session_messages, job_session_objects are spread across the indices even though the newest entries should probably be grouped together
Please consider doing a research project on how a switch to UUIDv7 will impact performance especially on medium to large multi-tenant installations
-
- Novice
- Posts: 6
- Liked: 7 times
- Joined: Aug 26, 2025 11:34 am
- Full Name: Torben
- Contact:
-
- Veeam Software
- Posts: 3664
- Liked: 877 times
- Joined: Oct 21, 2011 11:22 am
- Full Name: Polina Vasileva
- Contact:
Re: Btree indices and UUIDv7
Noted as well, as your previous feedback. I'm also passing this to our engineering team for review.
Thanks for sharing, and please keep it coming! )
Thanks for sharing, and please keep it coming! )
Who is online
Users browsing this forum: Bing [Bot] and 8 guests