Usage Metrics

We'll hear the term usage repeatedly when it comes to metrics. This is because there are various dimensions of usage metrics. In terms of infrastructure, we are measuring the usage of the resources of our API infrastructure.

In the following illustration, we can see the seven key infrastructure metrics that we must track across all our APIs.

In the following sections, we'll learn about these seven key infrastructure usage metrics in-depth.

Requests per minute

The RPM metric is a standard metric for gauging the application manager’s performance. Since the server doesn’t account for the latency caused by input and output (I/O) processes, it acts as a ceiling for an application’s interface.

RPM is commonly used to compare HTTP or database servers, as it is a metric of throughput. While some may take pride in a high number of RPM, it should be efficiency, not showmanship, that guides an engineering team’s efforts. While a high number of RPM might sound like a good thing, efficiency should drive an engineering team’s priorities.

To improve the RPM, it is necessary to design strategies that reduce the number of API calls made. We can reduce the number of API calls needed to perform certain business transactions to reduce the RPM. Having a flexible pagination mechanism and using common patterns such as batching many queries into one request may prove helpful.

If our API is designed for other businesses, we may notice that API traffic is lower on evenings and weekends, which means that our RPM reliance may change during different days of the week and times of the day.

CPU usage

CPU usage is an age-old indicator of performance that has been used for decades to manage resources and track patient wellness (a proxy for application responsiveness). High values mean the server/Virtual Machine (VM) is oversubscribed/overloaded or the ...