Key performance indicators give information about the application's responsiveness during normal behavior as well as under simulated workload. There are several indicators that can be gathered that directly affect the user, such as the response time or the error rate. These gauges represent the system's state and will provide insights about its behavior under performance tests. The indicated values will change depending on the number of concurrent users as well as the test scenario.

A first interesting insight is the application's response time - the time it takes to respond to a client's request including all transmissions. It directly affects the quality of the offered service. If the response time falls below a certain threshold, timeouts may occur that cancel and fail the request. The latency time is the time it takes until the server receives the first byte of the request. It mainly depends on the network setup.

During performance tests, it's especially interesting to see how the response time and latency time change compared to their average. When increasing the load on an application, at some point the application will become unresponsive. This unresponsiveness can originate from all kinds of reasons. For example, available connections or threads may be consumed, timeouts can occur, or database optimistic locking may fail. The request error rate represents the ratio of failed requests.

The number of concurrent users or load size in a specific interval of time affects the performance metrics of the application and needs to be considered in the test results. A higher number of users will put the system under more stress, depending on the nature of the request. This number is related to the number of concurrent transactions, technical transactions in this case, that indicate how many transactions the application can handle at a time.

The CPU and memory utilization provide insights about the application's resources. Whereas the values don't necessarily say much about the application's health, they represent the trend of resource consumption during load simulation.

Similarly, the overall throughput indicates the total amount of data that the server transmits to the connected users at any point in time.

The key performance indicators provide insights about the application's responsiveness. They help gather experience and especially trends during development. This experience can be used to verify future application versions. Especially after making changes in technology, implementation, or configuration, performance tests can indicate a potential performance impact.