I am writing a grant to provide students with free access to Nextcloud VMs, and I need to budget for physical machines. The question is, how can I estimate the number of VMs running just Nextcloud (with just the file sharing applications, basically--no videoconferencing or live editing or similar) that can be supported by a single physical machine? I need this to budget for the number of physical machines to purchase.
I have found some academic papers on this question, but I have no time to implement their models (and they are inadequately validated in real-world circumstances anyway). I have also seen some excessively rough methods of estimation (https://docs.citrix.com/en-us/tech-zone/design/design-decisions/single-server-scalability.html). These don't help much.
My approach can be rough, but it needs to be evidence-based. For instance, if you have experience with particular hardware supporting several Nextcloud VMs (or something similar to Nextcloud), that would be good. Or if you have thoughts about the factors that are most important to capacity estimation. I have considered load testing tools (Login VSI) as a way to validate my choices after the fact, but I don't see how I could do it to model things realistically in advance, so any thoughts there would be helpful too.
EDIT: The canonical answer Can you help me with my capacity planning? helps as a reference to include in the general lead up to the budget, but I need specifics for my case from someone with related experience. There's a chicken-and-egg problem: I can't first test some candidate machines and use the results to budget, since I would need money for the machines; but the money from the grant only comes in based on capacity planning. My hope is to base some of my estimates on the experiences of others to get around this problem.
