We’re currently deploying meteor on Ubuntu VMs on Azure, one for production and one for staging. We’re probably going to add US production/staging servers soon and will then assign a traffic manager to route to the best suitable VM per region. What I’m still wondering in terms of optimizations for our VMs is how we scale once we get more visitors. Do you usually create several VMs and they are connected to the same database, do you just upgrade the VM instance to use more RAM/cores (which shouldn’t matter with node.js afaik?) or how would we eventually have to change our setup when we scale?