Course contentsShow
← Java
Lesson 2781 of 2,87051. CompletableFutureFree lesson

Async Execution Models: Common vs Custom Executors

Understanding the default ForkJoinPool.commonPool() and when to provide a custom Executor for async tasks.

Async Execution Models: Common vs Custom Executors

You'll learn when CompletableFuture uses the default thread pool and when you should provide your own Executor.

The Default: ForkJoinPool.commonPool()

When you call async methods on CompletableFuture without specifying an Executor (like supplyAsync(supplier)), Java uses the ForkJoinPool.commonPool() by default. This is a shared, application-wide thread pool designed for CPU-intensive, parallelizable tasks.

Think of it like using the hotel's shared business center — convenient and zero setup, but everyone's fighting for the same resources.

When the Common Pool Works Fine

The common pool is perfectly adequate for:

  • CPU-bound tasks that don't block
  • Short-lived computations that complete quickly
  • Applications with minimal async workload
// Uses ForkJoinPool.commonPool() implicitly
CompletableFuture.supplyAsync(() -> expensiveComputation());

When You Need a Custom Executor

Provide your own Executor when:

  • Blocking I/O operations (file/network reads): The common pool's threads would sit idle, starving other tasks
  • Long-running tasks: You need dedicated threads that won't monopolize the common pool
  • Different thread priorities or configurations: Daemon status, thread names for debugging, custom thread factories
  • Isolation requirements: Separate pools for different subsystems prevent one component from blocking another
ExecutorService customPool = Executors.newFixedThreadPool(10);

// Explicitly use your custom pool
CompletableFuture.supplyAsync(() -> readFromDatabase(), customPool);

The async method variants ending in "Async" (like thenApplyAsync) accept an optional Executor parameter.

Key Takeaway:
Use the default common pool for quick, CPU-bound tasks; provide a custom Executor for blocking I/O, long-running operations, or when you need thread pool isolation and control.