Software web dev guy: Need to build on-prem test/stage/deploy chain

[LRP]

I’m building an ambitious Erlang web app. Need to think/act toward deployment. I’ve dug deep, get basic concepts, but need critique/mentoring of path forward.


Goal 1: Proxmox cluster to support testing, staging, and start-up production

Goal 2: Host production on-prem until start-up revenue supports move to cloud

Goal 3: Seamless, swiftest possible move to cloud when warranted


On-prem constraints: Low noise, low power, low-volume.


Resources: 2 NUCs, 4 MiniPCs, 1 Antsle server 4 Cores; at expense of marketing budget can add hardware resources within $1K to $2K range if necessary to achieve goal


Question 1: How best to harness existing resources under ProxMox to achieve

goal.

Question 2: How best to overcome one NIC per node constraint on existing hardware

Question 3: How best to provide secure storage to support VMs, Erlang Mnesia back-up and archiving (net 1 TB should be sufficient)

Question 4: How best to build out in stages toward goal


Many thanks to all,


LRP

 

[LnxBil]

I'd go with Docker and deploy to a PaaS not an IaaS, which PVE in fact is. You can setup e.g. GitLab with CI/CD pipelines as VMs, but I'd always deploy to Docker/Docker Swarm or Kubernetes. The later two can be VMs in your PVE, so you end up with 1 gitlab, 1 gitlab-runner and at least 1 VM for Docker and/or Orchestration. Every aspect can be clustered and be high available, but 3 is the minimum to start with.