In order to accurately model the relative differences in capability, Capacity Planner uses a concept of benchmark scores.
Each host segment in the Baseline View has a calculated industry standard benchmark, the SPECint rate. This benchmark data is produced by the Standard Performance Evaluation Corporation (SPEC). Capacity Planner normalises CPU utilisation on each machine to these SPECints. This is the equivalent utilisation expressed as a function of the chipset benchmark score.
To draw the sunburst accurately, the sunburst must have a reference machine from which to pin all sizing. It does this by automatically selecting the most common device type across the IT infrastructure. Each machine CPU capacity in SPECints is compared and drawn relative to that benchmark reference server.
When viewing CPU utilisation in the sunburst at a level higher than a host, for example, a cluster, data centre, application, or service, then that CPU utilisation is expressed relative to the benchmark reference server. This allows a consistent measure of CPU utilisation to be used across the whole estate when aggregating CPU values from various hardware platforms.
You can decide which host should be specified as reference. This host is then used to calculate normalised CPU demand for VMs in the Baseline View using these benchmarks, and establish relative performance or compute power.
Calculations for the headroom of a host include values taken from the reference host, the results of which are used in several aspects of the Baseline View. By default, the reference host is set as the most commonly available server in the current Baseline View.
To see the details of the reference host, follow these steps:
- From the configuration panel list on the left, click Benchmark.
The Benchmark panel opens with details about the current reference host.
- Click Locate server to locate the current reference host in the Baseline View.
The host is highlighted so that you can distinguish it from other hosts.