Maybe we need to let go of our auto-scaled 100 pod service mesh for a todo list app, and just deploy it bare metal on 2 servers.