Most of the "hardware maintainance" is defragmenting the workload so that the VMs run in fewer physical machines to retire racks of old hardware. Some load balancing too.
back
1 comments
Or an endless line of Spectre/Meltdown patches needing the hosts to reboot in order to get this weeks firmware into the intel CPUs.
This is what I was getting at. When Sun had issues with CPUs, they sent you new CPU boards and you could hot swap them live without downtime
That was indeed cool, but it also belongs to the "pet" phase of computing and this is the "cattle" age.
Not really. It was done at scale, too. Hotmail replaced all of the faulty CPUs live.