What is the difference between the cluster module and worker threads?

Both use more than one CPU core, but at different levels. A cluster worker is a full process with its own V8 heap, module state and event loop. A worker thread is a thread inside the same process with its own isolate but a shared process address space when you use SharedArrayBuffer.

import cluster from "node:cluster";
import { availableParallelism } from "node:os";

if (cluster.isPrimary) {
  const workers = availableParallelism();
  for (let index = 0; index < workers; index += 1) {
    cluster.fork();
  }
  cluster.on("exit", (worker) => {
    cluster.fork();
  });
} else {
  startHttpServer();
}

How to choose

• cluster for scaling an HTTP server across cores with process isolation
• worker_threads for CPU-heavy tasks that would otherwise block one server
• cluster workers do not share memory, so sessions and caches must be external
• worker threads can share memory but a crash can affect the whole process
• Containers usually make process-level scaling the platform's job, so cluster is less common inside Kubernetes

Sharing a port through cluster is convenient but the primary process still coordinates. In container platforms, running one process per container and scaling replicas is often simpler and gives better failure isolation. Use worker threads when the bottleneck is a specific CPU-bound operation, not when you simply need more request throughput.