It is worth reiterating that performance optimization begins with learning about how the underlying system works, and measuring the performance of systems we build under representative hardware and load. The chief component of performance optimization is identifying the bottlenecks using various kinds of measurements and profiling. Thereafter, we can apply experiments to tune the performance of code and measure/profile once again to verify. The tuning mechanism varies depending on the type of bottleneck.

In the next chapter, we will see how to address performance concerns when building applications. Our focus will be on the several common patterns that impact performance.