Benchmarking

It's difficult to compare the performance of algorithms simply by looking at their descriptions or pseudocode. It's better to run actual implementations on a real system. This method of measurement is called benchmarking. There are three categories of benchmark:

• Microbenchmarks: These are metrics showing the performance of certain functions. They assume that a small piece of business logic is contained in a single function, and we simply measure how fast this function runs.
• Macrobenchmarks: These are the opposite of microbenchmarks; they test the entire application.
• Mesobenchmarks: These are something in-between, measuring features or workflows.

Large applications more or less contain certain critical pieces of code. The JVM is an adaptive virtual machine, meaning it optimizes running code in many ways. Obtaining metrics that are meaningful is actually difficult, which is why microbenchmarks are the most interesting for developers.