@natts @NickFitz But the idea that you should auto-scale everything without bounds is farcical. There are other inputs to consider than raw uptime. Cost is a major driver and needs to be assessed vs the impact of downtime. Most will set a limit a bit above the max expected level