Performance issues after Blueprism 6.6 upgrade

[UdayB]

We have recently upgraded our Test Environment(PAT) to Blueprism 6.6.
After the upgrade, during the initial validation, we have noticed that the processes take an unusually long time to load inside the process studio.
A single process takes around 4.5 minutes to load, which will definitely have an impact on our developer productivity.

The same process when opened in another digital worker, on BP v 6.5 takes around 90 seconds, which is still too long, but something that was known. 
Further investigation shows the processor utilization peaking into 90% and the service becoming unresponsive for a period of time before coming back to normal. 

I wanted to check-in and see if any of the teams here has upgraded their environments to Blueprism v6.6 and faced any issues. Any pointers to the possible root cause for this performance degradation will also be very helpful. 

Thank you!


------------------------------
Uday B
Solution Designer

------------------------------

[david.l.morris]

I'll continue the off-topicness because I agree. Not only is the archiving feature not reliable enough, but it also doesn't archive the audit events log. Besides that, if the developers at BP have knowledge about best ways to handle the database such as max number of processes/objects or any other things like that, it may be worth committing that to paper. I feel like it's similar to the '100 RRs/ICs per Application Server' that is perhaps not an exact number but gives people a general guideline to work with. I'm guessing Ami is also aware of the lack of documentation in regard to maintaining the database. Maybe now that @Ami Barrett is at BP, this documentation might happen? Eh? =)​​

------------------------------
Dave Morris
3Ci @ Southern Company
Atlanta, GA
------------------------------

[AmiBarrett]

Oh yes. To both your points, I absolutely agree about the lack of documentation. However, there are multiple things that need to align in the background over here. My personal (completely unbased) theory, is that it leans toward what @VivekGoel was saying about the concern of people mucking with the DB. The re-indexing thing was something we stumbled upon at my last job after running a profiler to see why BPAWorkQueueItems was killing the box. There have been some optimizations here thankfully, but I do wonder if it's a similar/related issue. If you go that route, just be aware that it will temporarily lock the table and prevent anything from being worked - do it during a down period.

Anyway. I can't make any promises, especially given that I'm in a different division, but I'll see what I can push for. 😛

​Edit: Actually, some of this is already in a public document (link below). This includes mention of table reindexing, among other things.

https://portal.blueprism.com/system/files/2017-11/v6%20Data%20Sheet%20-%20Maintaining%20a%20Blue%20Prism%20Database%20Server.pdf

------------------------------
Ami Barrett
Sr Product Consultant
Blue Prism
Plano, TX
------------------------------

[Mustaheer]

We faced similar problem few days back, only BP was running slow rest of the applications were fine, then I requested BP help desk to share the housekeeping scripts for database (BP v6.6). after running those housekeeping scripts our systems are working super fast. I would suggest running housekeeping scripts on your db for your BP version.

------------------------------
Mustaheer Sayyed
Development Manager - IT
Rentokil PCI
Asia/Kolkata
------------------------------