- HA - for restarting failed VMs on good hosts
- vMotion - ability to migrate between different hosts
- DRS balances data storage and workload of datacenters
Reporting. If I lose a host, I have to report what was on that host, what was affected, and for how long. We create our own power state report providing info on hosts, and I would like reporting for what VMs were affected and what was the fault, what hosts were affected and restarted by HA. You can get third-party reporting, but it would be great to have it built-in.
Extremely stable, only issue is the type of hardware that we run it on. When we get a vendor image (HP, Dell, Cisco), they inject their own drivers into hypervisors, and it is only time we see issues. It’s so resilient, that these problems aren’t even evident and they can go unnoticted.
Very, very good scalability. Four years ago, we had 300 VMs across 32 hosts, now we have 2000 VMs across 132 hosts.
From time to time, we have relied on them, mostly for troublshooeting specific VMs when a vendor wants to blame resources. Whenever we have bad memories or drivers injected by vendors, they’ve helped us remediate those issues.
I was involved in a redesign, which was very straightforward and easy. We just changed the hardware from rack style to blade, expanding all hosts and underlying virtualization layers. It was seemless and customers didn’t notice anything different.
Even with no reporting, it's near perfect, because reporting would just make things easier, and not having it doesn’t impede performance.