System Performance is about more than you think
What is the common view of system performance metrics?
System performance is commonly regarded as processing capacity and response times, however, system performance covers most of the non-functional aspects of an IT system's behaviour, such as system availability, system robustness and recovery, service production efficiency and an almost endless number of more metrics.
Are all metrics of interest?
The simple answer is YES. They all contribute in different ways to an in-depth understanding of how the system behaves in different situations. This is needed to proactively identify potential performance issues to avoid annoying or even catastrophic issues before customers are affected. Furthermore characteristics are used to correctly configure a system for a specified workload.
It is also needed to identify a system's design limitations for future or immediate actions. System performance measurements are a learning process about a system's behaviour and that is what makes it a very interesting topic.
What makes system performance measurements more complicated?
What is common to all performance metrics is that the measurement results are evaluated on a scale from excellent to very poor unlike functionality test cases where test results are binary: passed or not passed. It also makes regression testing more complicated. System performance metrics must be compared to prevoius result to see if results are better, the same, or worse than for last measurement, i.e. a performance history database is required. Unlike functionality requirements system performance requirements can always be improved.
What is it more than capacity and response time?
The following pages are a short introduction in system performance and we hope reading them will bring you an interest in the topic and the importance of system performance. In the first edition we cover the following basic concepts:
More pages will be added to this section every month during 2019.
Next addition will be about Resources and Performance measurements.