Service allocation approach?

[Qacer]

I'm looking for some tips on how to approach optimizing the allocation of field service agents. Basically, field service agents are experts who are able to troubleshoot a system that are installed in different regions. They have to be present in order to troubleshoot a system. The challenge is that there are more systems than field service agents. Also, the number of systems that will be installed grows at a higher rate compared to the number of field service agents hired. One of the main constraints is that the plan to hire a certain amount of service agents and at certain locations have to be submitted and approve two years before the actual hiring, so the right data need to be captured properly and consistently to do trending reports. Another constraint is that demand surges are almost always expected, but cannot be properly predicted despite trending reports. Some form of contingency or backup plan need to be factored in order to have a sufficient amount of agents at an acceptable risk.

Anyway, I'm looking for best practices, tips, etc. on how to approach this problem. If you have any real life example, then that would be great. I'm also curious what types of data you factored in, what type of analysis you conducted, processes adapted, etc.

Thanks!