Go Cookbook

Chapter 19. Benchmarking Recipes

19.0 Introduction

Performance testing is an important part of software testing. It’s nonfunctional, meaning it doesn’t test whether the software does what it’s supposed to do. Instead, it tests how well the software performs in terms of stability, speed, and scalability under a given workload.

In Go, performance testing is done with the same testing tools as functional testing, using the go test tool and the testing standard library.

19.1 Automating Performance Tests

Problem

You want to automate the performance testing of a function.

Solution

Create a benchmark function and use the go test tool to run it.

Discussion

Go provides a minimalist set of built-in tools for testing with the go test command-line tool and the testing package. These tools are used for both functional and performance testing. In this recipe, we’ll focus on performance testing.

As in Chapter 18, you’ll be using a simple Add function and try to do performance testing on it:

package test

func Add(a, b int) int {
	return a + b
}

The test flow for performance testing is the same as functional testing. First you define a set of benchmark functions in a file that has a general naming convention of _test.go. In this case, put your benchmark functions in a benchmark_test.go file. As in functional tests, the name of the package the file is in must be the same as the one with the function you want to do performance testing on:

func BenchmarkAdd(b *testing.B) {
	for i := 0;