fbpx

Why does NodeJS scale insanely?

If you are new to NodeJS, you might have heard that NodeJS is single threaded. And you might have also heard that it is insanely scalable, serving millions of users at realtime. How does a single threaded application scale so well?

Single threading is half the truth

Yes, NodeJS follows a Single Thread Event Loop Model, but it is not actually single threaded. It works on an event based execution architecture.

NodeJS has a main thread and additional worker threads. Tasks that do not have to be serviced synchronously can be passed onto the worker threads. When worker threats are ready to be executed, they report back to the event loop. The event loop picks up an event and passes it to the main program stack for being the next in line for execution.

This provides a single threaded, but sudo parallel execution environment.

Understanding NodeJS Execution

const request = require('request');
let f1 = function() {
  console.log('Hello at beginning');
  request('https://google.com', (err, res, body) => {
    console.log('Hello from function');
  });
  console.log('Hello at end');
}

f1();

If we executed the above code in a procedural manner, we would expect the following output.

Hello at beginning
Hello from function
Hello at end

However your NodeJS application will show the following output.

Hello at beginning
Hello at end
Hello from function

Why is this so? Why does the request() line execute after the last console.log() statement? This is so because invoking request() is an asynchronous task. The execution of this task gets allotted to a worker thread. While the worker thread waits to get the response from google.com, the main thread can continue with further execution. This results in the last console output being printed while the worker thread is waiting for a response on the request.

When the worker thread does receive a response, it puts an entry into the event loop. When the main thread is free and doing nothing else, it picks up an event from the event loop and executes the tasks that was allotted to the worker. The event loop tasks are only executed when the main thread is free and not performing any other task.

NodeJS Async Execution
Call to request() passed on to a worker thread

So why is NodeJS insanely scalable?

This unique event based model prevents NodeJS from being blocked by any specific event. Each event is treated and processed independent of each other. This is only true as long as you don’t write code that blocks the main event thread.

Since async function calls report back to an event loop for execution when they are ready to be executed, the main thread is always busy doing something and never waiting on any task. A properly designed NodeJS application, can thereby keep the main event loop free from long running tasks, by passing long running tasks to worker threads.

This concept is very different than spawning new threads for executing tasks in parallel. There is a physical limit to the number of threads a system can execute. When this limit is reached, if individual threads are waiting for a long running operation to complete, all threads would essentially wait, thereby making the complete application slow.

On the contrary, in NodeJS, the main event loop only gets those tasks to execute that are ready to be executed. Thereby millions of concurrent events can be created, without affecting the performance of the main thread, thereby allowing for significant scalability of applications that are well designed.

NodeJS is turning out to be one of the preferred backend systems for web applications and web services.

%d bloggers like this: