Our cluster has been running overprovisioned by design for years. The state of mind was to avoid incidents and deal with the bill later. now I'm being asked to deal with the bill and I have no good options.
manual optimization is out of the question as we're way past the scale where that's realistic. and management is wary of designing our own automation because of the engineering hours it will cost for development, implementation, and troubleshooting.
so that leaves me with the option of getting a new automated tool. which I really don't want. we got burned badly a few years ago. since then it's been a hard no for me.
Am I missing a fourth option? and if automation really is the only way to go, is it actually safe now? because even if I am being forced to get a tool I dont want, I will still be the one who gets blamed if it goes sideways.