Now that we've implemented a REST-based server, we can return to the Fibonacci application, applying what we've learned to improve it. We will lift some of the code from fiboclient.js and transplant it into the application to do this. Create a new file, routes/fibonacci-rest.js, with the following code:

const express = require('express');
const router = express.Router();
const http = require('http');
const math = require('../math');

router.get('/', function(req, res, next) {
  if (req.query.fibonum) {
    var httpreq = http.request({ 
      host: "localhost", 
      port: process.env.SERVERPORT, 
      path: "/fibonacci/"+Math.floor(req.query.fibonum), 
      method: 'GET' 
    });
    httpreq.on('response', response => {
      response.on('data', chunk => { 
        var data = JSON.parse(chunk); 
        res.render('fibonacci', { 
          title: "Calculate Fibonacci numbers", 
          fibonum: req.query.fibonum, 
          fiboval: data.result 
        }); 
      }); 
      response.on('error', err => { next(err); }); 
    });
    httpreq.on('error', err => { next(err); }); 
    httpreq.end(); 
  } else {
    res.render('fibonacci', {
      title: "Calculate Fibonacci numbers",
      fiboval: undefined
    });
  }
});

module.exports = router;

In app.js, make this change:

const index = require('./routes/index');
// const fibonacci = require('./routes/fibonacci');
// const fibonacci = require('./routes/fibonacci-async1');
// const fibonacci = require('./routes/fibonacci-await');
const fibonacci = require('./routes/fibonacci-rest');

Then, in package.json, change the scripts entry to the following:

"scripts": {
  "start": "DEBUG=fibonacci:* node ./bin/www",
  "startrest": "DEBUG=fibonacci:* SERVERPORT=3002 node ./bin/www",
  "server": "DEBUG=fibonacci:* SERVERPORT=3002 node ./fiboserver" ,
  "client": "DEBUG=fibonacci:* SERVERPORT=3002 node ./fiboclient" 
},

How can we have the same value for SERVERPORT for all three scripts entries?  The answer is that the variable is used differently in different places. In startrest, that variable is used in routes/fibonacci-rest.js to know at which port the REST service is running. Likewise, in client, fiboclient.js uses that variable for the same purpose. Finally, in server, the fiboserver.js script uses the SERVERPORT variable to know which port to listen on.

In start and startrest, no value is given for PORT. In both cases, bin/www defaults to PORT=3000 if it is not specified.

In one command window, start the backend server, and in the other, start the application. Open a browser window as before, and make a few requests. You should see output similar to this:

$ npm run server

> fibonacci@0.0.0 server /Users/David/chap04/fibonacci
> DEBUG=fibonacci:* SERVERPORT=3002 node ./fiboserver

GET /fibonacci/34 200 21124.036 ms - 27
GET /fibonacci/12 200 1.578 ms - 23
GET /fibonacci/16 200 6.600 ms - 23
GET /fibonacci/20 200 33.980 ms - 24
GET /fibonacci/28 200 1257.514 ms - 26

The output like this for the application:

$ npm run startrest

> fibonacci@0.0.0 startrest /Users/David/chap04/fibonacci
> DEBUG=fibonacci:* SERVERPORT=3002 node ./bin/www

  fibonacci:server Listening on port 3000 +0ms
GET /fibonacci?fibonum=34 200 21317.792 ms - 548
GET /stylesheets/style.css 304 20.952 ms - -
GET /fibonacci?fibonum=12 304 109.516 ms - -
GET /stylesheets/style.css 304 0.465 ms - -
GET /fibonacci?fibonum=16 200 83.067 ms - 544
GET /stylesheets/style.css 304 0.900 ms - -
GET /fibonacci?fibonum=20 200 221.842 ms - 545
GET /stylesheets/style.css 304 0.778 ms - -
GET /fibonacci?fibonum=28 200 1428.292 ms - 547
GET /stylesheets/style.css 304 19.083 ms - -

Because we haven't changed the templates, the screen will look exactly as it did earlier.

We may run into another problem with this solution. The asynchronous implementation of our inefficient Fibonacci algorithm may cause the Fibonacci service process to run out of memory. In the Node.js FAQ, https://github.com/nodejs/node/wiki/FAQ, it's suggested to use the --max_old_space_size flag. You'd add this in package.json as follows:

"server": "SERVERPORT=3002 node ./fiboserver --max_old_space_size 5000",

However, the FAQ also says that if you're running into maximum memory space problems, your application should probably be refactored. This gets back to our point several pages ago that there are several approaches to addressing performance problems, one of which is the algorithmic refactoring of your application.

Why go to the trouble of developing this REST server when we could just directly use fibonacciAsync?

We can now push the CPU load for this heavyweight calculation to a separate server. Doing so would preserve CPU capacity on the frontend server so it can attend to web browsers. GPU co-processors are now widely used for numerical computing and can be accessed via a simple network API. The heavy computation can be kept separate, and you can even deploy a cluster of backend servers sitting behind a load balancer, evenly distributing requests.

What we've demonstrated is that it's possible to implement simple multitier REST services in a few lines of Node.js and Express.