I'm not an expert, but what I see differs from that of Richardson. Assuming I am looking at resources from the level of a cluster, what I see is that in your policy you have set a 10% buffer on CPU demand, on a cluster which has 10GHz of CPU. Obviously, you have 9GHz before breaching the threshold you have set in the policy. I can also see that at peak demand, you are running at 124%, so you must have spikes and peaks enabled in your policy as well.
The reason you have 0% left, is that the peak demand exceeds the 9GHz you have told vROps you should have available. (Even then you are above 10GHz, so even without the buffer you would be at 0%).
For comparison purposes, I would uncheck the option to measure against spikes and peaks and see what that leaves you. You can also evaluate the CPU Demand (or CPU Workload) at the cluster layer to see where these spikes are to look into possible level setting them. Backups, AV, etc, are usually the middle of the night culprits which drive up peak demand.
All of that being said, in the end having spikes and peaks enabled is definitely the better way to go. You want to have capacity to accommodate those peak times. Only uncheck it for a capacity comparison to what you have now, and as I mentioned, see if you can spread out things like backups if you find that the spikes are much higher than normal operating average.
If however you are running "hot" all of the time, (again, based on the policies you have set), then you may want to look into adding more physical resources, not more virtual.
If you find this or any other answer useful please mark the answer as correct or helpful
https://communities.vmware.com/people/greco827/blog
