With performance monitoring, you can monitor how well your solution performs under different circumstances. During the development of a solution, there probably won't be a high load on the components, so it makes sense to test how the solution behaves under high load and, it's been once brought to production, to monitor that performance. You could monitor, for example, any of the following:
- Processed transactions
- Transactions per second
- Requests per second
- User satisfaction
You can measure these metrics per component, but also for the entire chain. By segregating per component, it will be possible to identify any bottlenecks. Also, if your solution has to cope with drastically changing workloads, it is good to know that, depending on the Azure component in question, it is easy to scale up there is a heavy load and to scale down when under an average load. In this way, you only pay for resources when you need them, which can be handy if you expect more load on your solution during particular periods of the year.
Another aspect that falls under the category of performance monitoring is user satisfaction. To measure this, you can use Apdex scores (https://www.apdex.org). Apdex scores scale from zero (no users satisfied) to 1 (all users satisfied). Performance results are divided into the following categories:
- Satisfied
- Tolerating
- Frustrated
You can find the Apdex score specification here: http://www.apdex.org/index.php/alliance/specifications/.