Thread Pool & Executor Framework

A thread pool is a collection of pre-created reusable threads managed by ExecutorService, providing bounded concurrency, task queueing, thread reuse, and lifecycle management.

The Executor framework decouples task submission from task execution: Executor provides fire-and-forget execute(Runnable), ExecutorService adds submit(Callable) returning Future for result retrieval, lifecycle management, and batch operations.

Runnable defines void run() with no return value and no checked exceptions, while Callable<V> defines V call() throws Exception, returning a result through a Future<V> when submitted to an ExecutorService.

With execute(), an unchecked exception kills the worker thread and goes to UncaughtExceptionHandler; with submit(), the exception is captured inside the Future and silently lost if get() is never called.

When InterruptedException is caught, the interrupt flag is automatically cleared; always restore it with Thread.currentThread().interrupt() or propagate it, because swallowing it breaks shutdown, cancellation, and graceful termination.

ThreadPoolExecutor takes 7 parameters (core, max, keepAlive, unit, queue, factory, handler) and follows the flow: core workers first, then queue, then max workers, then rejection policy.

Unbounded queues cause OOM and latency tail disasters because tasks pile up forever; bounded queues provide predictable fail-fast rejection; production systems need custom handlers with metrics, logging, and back-pressure.

Five BlockingQueue implementations serve different patterns: ArrayBQ (fixed, fair), LinkedBQ (two locks, high throughput), PriorityBQ (heap-ordered), SynchronousQueue (zero capacity, direct handoff), and DelayQueue (time-based release).