Parallel Flow in Node Modules

Question

Parallel Flow in Node Modules

I am having an express app. I want to have parallel flow for an array of functions that I want to run. I am thinking of using async module for doing so.

I want to know if there is any other module which will be more better than this?

Secondly I want to know is it necessary that these functions be Asynchronous? Lets say I have code like this

var sum = function(x, y){
   return (x + y)
}

async.parallel([
    function(callback){
        setTimeout(function(){
            result = sum (x, y); //just an example for a synchronous function
            callback(null, result);
        }, 100);
    },
    function(callback){
        result = sum (x, y); //just an example for a synchronous function
        callback(null, result);
    }

],
// optional callback
function(err, results){
    console.log(result);
    // the results array will equal ['one','two'] even though
    // the second function had a shorter timeout.
});

SO as you can inside this there are functions which are synchronous. So will still this two run in parallel?

I have also heard that in node.js only I/O tasks can run in parallel as node.js is single threaded. Is it true? And so if I don't have I/O tasks that in async module also they won't run in parallel rather just appear to?

Please help.

node.js
asynchronous
express
node-async

Answer 1

Node.js is single-threaded and (out of the box) single-core.

So, nothing is truly parallel in Node.js since is single-threaded.

The async.parallel method is useful only if your tasks contain asynchronous code: it will wait for all the responses, then executes the code in the callback. But, if your tasks contains only synchronous code, the result will be synchronous.

So the answer to your last question is yes.

If you want to try another control flow module, try Q. It implements Javascript Promises and makes async code easier to read and to organize.

If you want to truly parallelize your tasks, look at child process or cluster module.

There are also some special native modules as Threads à gogo which implements threading.

Full example with your code:

var async = require('async');

sum = function(x, y) {
  return (x + y)
}

x = 2;
y = 3;

async.parallel([
    function(callback) {
      setTimeout(function() {
        result = sum(x, y);
        console.log('f1');
        callback(null, result);
      }, 100);
    },
    function(callback) {
      result = sum(x, y);
      console.log('f2');
      callback(null, result);
    }

  ],

  function(err, results) {
    console.log(result);
  }
);

If you run this piece of code output result will always be: f2, f1, 5. This is because of setTimeout in the first function, but the way the code is executed is synchronous.

If you want to be convinced, remove setTimeout in the first function and see that output will always be f1, f2, 5.

So:

async.parallel is very useful if you want to run two HTTP requests (which are asynchronous) at the same time, and wait for both of them to be completed.
async.parallel is not useful if you put synchronous code in it, because the code will be executed in a synchronous way.

Answer 2

async is the de facto module for these kind of tasks, go with it.

No, it is not mandatory for those function to be asynchronous. I belive all your questions could be answered with this: http://blog.mixu.net/2011/02/01/understanding-the-node-js-event-loop/, where the following is the important point:

Node.js keeps a single thread for your code however, everything runs in parallel except your code.