Modern Java Recipes

Chapter 9. Parallelism and Concurrency

This chapter confronts the issues of parallelization and concurrency in Java 8. Some of the concepts extend back to language additions from much earlier versions of the language (especially the java.util.concurrent package added in Java 5), but Java 8 specifically added several capabilities to the language to help you operate at a higher level of abstraction.

One hazard of parallelization and concurrency is that when you try to talk about them, someone will care a lot—very vocally—about the distinction between the two words. Let’s get that out of the way right now:

• Concurrency is when multiple tasks can run in overlapping time periods

• Parallelism is when multiple tasks run at literally the same time

You design for concurrency—the ability to decompose your problem into independent operations that can run simultaneously, even if they aren’t doing so at the moment. A concurrent application is composed of independently executing processes. You can then implement the concurrent tasks in parallel, which may or may not improve performance, assuming you have multiple processing units.¹

Why wouldn’t parallelization help performance? There are many reasons, but parallelization in Java by default splits work into multiple sections, assigning each to the common fork-join pool, executing them, and joining the results together. All of that work introduces overhead. A lot of expected performance improvements will be decided by how well your problem ...