Raw Capacity Required for Workloads
To set the stage for the remainder of the example, let’s identify a desired workload. The customer wants to host these many VMs:
-
1000 small = 1000 vCPU, 4000GiB. Plus 228 GiB memory overhead.
-
300 medium = 600 vCPU, 2400GiB. Plus 73 GiB memory overhead.
-
200 large = 800 vCPU, 4800GiB. Plus 58 GiB memory overhead.
-
200 x-large = 1600 vCPU, 9600GiB. Plus 64 GiB memory overhead.
This brings us to a total of 1700 VMs using 4000 vCPUs and 20800 GiB memory, plus 423 GiB overhead. The average VM will have 2.4 vCPUs and use 12.5 GiB memory, which is a convenient way of estimating capacity for nodes.