think of capacity as being how much of your current azure quotas have you used up. if you asked for 100 cores, and you are using 68 of them (for whatever) then that will show 68% capacity used.
(its a little more complicated that this, which i will explain later). Some (not necessarily all) of the capacity is being used to provide the Horizon Cloud Service for end users to use. (eg for the underlying node/uag appliances, along with any servers that you provision for farms etc)
Utilization is now talking about how much of your Horizon Cloud Service deployed servers as RDS Farms are actually being used for end users (and how much).
Lets say you had a Farm with 1 server, with Max sessions per server = 20. and you have 10 users using it, then 10/(1*20) == 50% utilized. ie your end users are using 50% of the provided capacity.
if you had a farm with 10 servers in it, and 20 sessions per server, with just 10 users using it, then the utilization would be 10/(10*20) == 5% utilized.
during the day as users login, and use the service then the utilization will increase, and as an administrator you should look to make sure that it doesn't hit (or get close to?) 100%.
If it does get close to 100%, then assuming you still have some capacity headroom, then you can make the farm max size bigger to accommodate.
In most cases, if your capacity gets close to 100% then you can request an increase in cores via the azure portal subject to any hard limits that Microsoft impose.
Note that utilization considers 'max' servers in a farm, and isn't worried that some number of those servers may be powered off. powered off servers are still considered part of the utilization calculation denominator.
For capacity (I said it was a little harder than I first presented); Horizon Cloud Service uses a number of different vm sizes. We also dont have a dedicated usage of your Azure subscription. you can deploy other vm's etc into the environment. From the Azure portal (Usage and Quotas) you can see something like this (I just snapped this from one of my test subscriptions):
from this, you can see they are reporting manu different things. there isnt a single 'usage % number'.
What we do in Horizon Cloud Service is to consider the 'worst case' for the items we need. (ie we dont worry about other azure vm families that you may have deployed etc), but for our service we know that you (currently) use Dv2 family of VM's for RDS Servers. We also use F series for jumpbox, and Av2 for UAG appliances. We need family cores, and regional cores for the node, and we need 'vms'. So, our calculations look across all these values, and then we return as that top level number the 'worst case' %used out of them, and report that as that top level number.
We have a UI that actually does a pretty good job at helping you see how this is calculated. If you click Settings->Capacity, then view by 'Type' , next to the subscription name (which incidentally you can click to update the azure service principle - eg to rotate out your secret) there is a progress bar. (yes, thats a little obscure, and we are fixing that shortly!), anyway, click the progress bar, and a UI like this will be displayed:
this in my case shows that the 'worst' item for eastus region for me is cores, and thats at 65%. So for me, my overall reported capacity is 65%. I have plenty of headroom on my DV2 family cores (32% used) etc.... The first thing thats going to run out here, and prevent me expanding my deployment will in my case be cores. It may well be different for you.
Anyway, I hope that explains in detail the meaning of capacity and utilization, and you have a better understanding of how the capacity is calculated and where to go see that,
im definitely keeping you busy! thanks again.