Definition: Benchmark testing measures a system's performance against a fixed, known reference point, such as a set data set, workload or previous result. Running the same benchmark over time shows whether performance is improving, staying steady or getting worse.
So what? Without a stable benchmark, performance results are hard to compare between builds. A repeatable benchmark lets teams spot performance regressions quickly, even within a short iteration.
Examples: Running the same set of queries against a static, read-only installation after each release. Comparing response times for a known data set before and after a configuration change.
