That's pretty much the reporting solution I'm trying to replace with performance analytics. With that many assignment groups and several million rows in the incident table it's catastrophically slow so I want to do the heavy lifting overnight rather than on-demand.
My initial attempt had a breakdown source of all problem records that have had an incident which was created in the previous month related to them, then apply that as a breakdown to an indicator that counts the number of incidents linked to problems in the previous month. That seemed to work ok but was problematic when I tried to actually use it in a breakdown widget - I got no data to display despite being able to see in the analytics hub the collection had happened. However, with c1500 assignment groups and c2000 problems I managed to kill the performance of the analytics hub too, and as the point is to fix the performance issues of the existing solution that's not ideal!
The primary use of the data is as part of a risk control - teams are required on a monthly basis to make sure their most active moderate priority problems (currently that's based on number of related incidents in the previous month, NOT total number of related incidents) have some sort of status update.
I think I'm going to investigate setting up a separate spotlight solution using an indicator and see how that works, that might be a better way of going about this. It would also give me more control over criteria - if we want to give more priority to certain types of problem in response to user feedback we can just tweak the spotlight scores.
