Improve Scalability of PI Analytics
Currently you can only run a single PI Analytics instance at a time. We have two VMs running PI Analytics in a failover situation, but this means that one of the VMs is sitting idle 99% of the time.
We are starting to hit the limits of scaling with analytics falling behind. Ideally, the analytics should be able to run on multiple VMs.
For example, if you have 10 AF Databases and 2 Analytics servers, it should run the analytics for 5 of the databases on each VM. In a failover situation it could send the Analytics for all 10 databases to the online server. In this way if you run into performance issues you could increase the number of Analytics servers to improve performance (at least across multiple AF Databases). If you added a 3rd server, it would split the jobs 3+3+4 based on AF Databases.
You would still need to ensure the servers are adequately specified to handle the load of a single AF databases calculations, but this would go a long way to improving the problems we are having with performance.