Set available processors in Executors.newFixedThreadPool in HTTPServer

Question

I'm building an HTTPServer that needs to handle multiple requests at the same time.

The main function of what I built looks like this:

public static void main(String[] args) throws Exception {
    HttpServer server = HttpServer.create(new InetSocketAddress(8000), 0);
    server.createContext("/", new MyRequestDispatcher());

    server.setExecutor(Executors.newCachedThreadPool());
    server.start();
}

Now I'm thinking on how this Executors.newCachedThreadPool() works regarding the number of threads created. As I have read that the number of threads to be created is not limited, if I get a thousand of requests at the same time, would it create a thousand of threads?

I'm thinking on limiting the number of threads created at the same time in order to be handled properly in the machine where it is being run. I thought on something like this:

Executors.newFixedThreadPool(Runtime.getRuntime().availableProcessors())

The goal is to only create the given number of threads depending on the available processors in the system.

Would this work?

Thanks in advance!!

Answer 1

Yes, it will work, and it's what you'll see recommended most often.

Depending on your specific use-case you still may want to use a different number.

• If your tasks are blocking (e.g. because of I/O operations), you may benefit from more threads than available cores.
• If you can run multiple hardware threads per core, you may also benefit from more threads than available cores. For example, Intel x86 Xeons with Hyperthreading enabled can run 2 threads per core, IBM POWER8 and Oracle SPARC M7 - up to 8 threads per core.
• If you are running concurrent GC collectors, you experience significant GC pressure, and you have a non-stop running thread pool, you may find that slightly decreasing the number of threads in the pool can move you in a sweeter spot. This is a fairly specific example, though, which I'm mentioning because it was a big revelation in the system I'm currently working on.