Performance lags on large database

Hello,

I have some information concerning areas to target for possible performance optimizations, that I hope can help in view of the upcoming work for v9.2.
Our tests were conducted on v9.0.6, using a large database set.

Test scenario

1. Login as Project leader (affected to only 5 small projects)
   
2. Create new project '0002-01-autre lot1', at root of project structure
   => Save operation is quick
    
   
   
   
3. Update project '0002-01-autre lot1', set "is sub-project of" = '0002-Test'
   => Save operation is very long, and log traces show a huge number of projects are saved.

It looks like there's something wrong in the recursive save operation, because there are no dependencies apart from the parent project and the sub-project.
Can this help to figure out what could be improved ?

Timestamps in the log file :
- First save operation : 2021-04-20 15:46:46.640 => 0.000

---

DEBUG

---

Start SAVE for Project #
- Second save operation : 2021-04-20 15:47:27.499 => 0.000

---

DEBUG

---

Start SAVE for Project #16873


 

That's what I was planning
I'll keep you posted

Hi,

I've tested the same scenario after upgrading to v9.1.2, but I am still seeing the same behaviour : when moving the new project to "is sub-project of" parent project, a huge number of other projects are also saved.
I checked the sortOrder, which looks ok and consistent between 'project' and 'planningelement' tables.  (In this case it's project #16877 that I'm moving to sub-project of #16872).

 

Would you see anything else that could explain this ?