Wrapping Up

Elixir provides a wide range of tools to help you instrument your application to measure performance in production. The first section showed how to monitor and fine-tune Observer to ensure everything runs well in production. The Erlang Observer tool uses the same low-level APIs to collect data that you can use to instrument your own solutions. You learned how to use those APIs to measure memory usage, resources by process, and other details such as run queue length.

The next topic was profiling. You can use three important tools to count function call frequencies or call time (or both), and each has inherent trade-offs. Profiling can give you a good idea of where a performance problem might be, down to the offending function.

Finally, we ran a benchmark to fix a problem. Our code benchmarked a potential solution against the offending code to verify that the cure was in fact better than the disease.

Since you now know how to monitor an application that’s running well, it’s time to shift gears. In the next chapter, you’ll learn ways to be aware of application errors and what we can do about it. There’s one more chapter to go, so let’s dig in!