Enhancements to the dashboard
The preceding dashboards should be sufficient. However, if you think you need additional visibility to help you understand your environment, you can:
- Use the percentile chart
- Add utilization information
We will use compute as an example. It should be relatively straightforward to apply the idea to storage and network.
The percentile chart
The maximum value can be too aggressive. All it takes is one VM experiencing high contention, and you have to consider the cluster full already. This is fine if your IaaS is indeed unable to meet the business demand. What if the demand is not from the business but IT's own generated workload?
An example is a full AV scan or full backup that was performed during non-business hours. They may impact the VMs, making them suffer from contention. You may consider that acceptable if business is not affected.
This is where standard deviation and percentile come in. You can find out the value at, say, the 95th percentile, giving you...