virtual machine scale sets / managed instance groups / auto scaling group support

[colemickens]

Hello. The major cloud providers expose higher-level constructs above VMs that represent (largely) homogenous groups of machines that are defined by a single logical unit in the cloud provider control plane.

• Azure: Virtual Machine Scale Sets
• AWS: Auto Scaling Groups
• GCP: Managed Instance Groups

I couldn't find anything in existing issues, but has any consideration been given to how they might be exposed in NixOps? I haven't had a chance to dig into the internals yet (soon, hopefully) but I'm curious if this has been discussed before. Is it specifically in or out of scope for NixOps?

There are some use cases that make NixOS and NixOps extremely attractive, particular with large scales of machines. Certain cloud providers might not work particularly well with the single-VM constructs at certain scales, and expect users to be on the multi-homogenous-VM construct for large clusters of VMs, so I'm interested in seeing if NixOps could understand these constructs.