Once we get insight into the code via testing and profiling results, we need to analyze the bottlenecks worth considering for optimization. A better approach is to find the most under-performing portion and optimize it, thereby eliminating the weakest link. We discussed performance aspects of hardware and JVM/Clojure in previous chapters. Optimization and tuning requires rethinking the design and code in light of those aspects, and then refactoring for performance objectives.

Once we establish the performance bottlenecks, we have to pinpoint the root cause and experiment with improvisations, one step at a time, to see what works. Tuning for performance is an iterative process that is backed by measurement, monitoring and experimentation.