🐴 Distribute load on front-end via parallelism
Switch branches/tags
Clone or download

README.md

super-workers

Super Workers
Distribute load on front-end via parallelism
A 5.5kb JavaScript library for executing multiple parallel tasks on front-end


npm version npm Build Status Coverage Status Donate

Table of Contents

Features

  1. Run multiple tasks parallelly without blocking the main thread.
  2. Jank-free pages since all the computations take place in worker-thread and not in main-thread (thereby speed and performance at the same time).
  3. On-demand initialization of web-workers based on task-queue size. Configure minimum and a maximum number of web-workers or let library smartly decide based on hardware concurrency number.
  4. Configure task(set priority, time-to-live) and Priority Task Queue Management System(PTQMS) will take care of the queue handling and execution.
  5. Termination of idle(non-used) web-workers for better performance(can be opt-out while configuring). Set minimum number of web-workers to restrict their termination when idle, so that they remain present for future needs.
  6. Fully fledged API to get all the information regarding tasks and workers at any given point of time.
  7. Promise based execution of tasks i.e. perform different actions on success or failure. Useful for the re-attempting failed task.
  8. Define callbacks to be get triggered at various events. Eg. on the successful execution of the task, on exceeding the time limit of task execution, etc.
  9. Exports in a UMD format i.e. library works everywhere (nodejs doesn't have the concept of workers but have a child-process concept which is not handled in this lib).
  10. Only ~5.5 KB gzipped.

Installation

  1. Via npm
$ npm install super-workers
  1. Via yarn
$ yarn add super-workers
  1. Via bower
$ bower install super-workers
  1. Download source - Link

Flow Diagram

Aim - To be able to run multiple tasks simultaneously without freezing the web-page

  • Tasks are queued as they arrive.
  • Task Queue Process Management(TQPM) processes the queue and based on tasks' priority, it reorders the queue.
  • Workers are spawned as and when required based on number of tasks and configuration.
  • Availability of workers decide the number of tasks getting processed.
  • Completion of a task allow the next queued task to be get executed.

Below diagram will aid in understanding the flow better


Explanation of diagram

Setting the workers configuration

Lets have the following worker configuration:

  • minimum number of workers always available to execute tasks - minWorkers = 1
  • maximum number of workers that should be spwaned irrespective of the number of tasks in the queue - maxWorkers = 3. This number will restrict registering of workers more than this number. We can not set this number randomly since it depend on the number of cores in a machine. By default, the library will pick it for you using navigator.hardwareConcurrency or it's polyfill. Setting it too high will not boost queue processing as it becomes constant after the threshold(totally depends upon the hardware configuration).

Task Queue Process Management

  • Initially the task queue(TQ) is empty.
  • Task T1, T2, and T3 arrive in the same order but with different priorities.
  • Task T1 is of medium priority.
  • T1 is pushed into queue and Task Queue Process Management reorders the TQ based on tasks' prioriies.
  • After proper ordering of tasks, task is ready to be executed by the web-worker

Web worker Execution

  • Firstly, availability of workers is checked.
  • If there is no worker available, there are two cases. i) All workers are busy i.e. maxWorkers number has been reached, so, tasks need to be in queue for more time before they can be picked up for execution. ii) No worker has been spwaned yet and maxNumbers is greater than the number of spawned workers.
  • If the condition for spwaning a new worker passes, a web-worker is registered and the task is removed from the TQ and is executed.
  • Worker(s) execute(s) the task(s) and return a promise.
  • After completion of task, the process is re-run to execute next pending tasks in the TQ.

Usage

Create an instance / reference before using.

Main Thread

var sp = new SuperWorkers.MainThread({
    url: 'js/my-worker.js'
});

Constructor accepts a configurable Object with the following keys.

  • url: The path of worker script
  • maxWorkers: Maximum workers that can be spawned to execute parallel tasks. Depends on CPU cores(hardwareConcurrency) of a machine
  • minWorkers: Minimum workers that should be always available to execute parallel tasks (one worker should alive if tasks keep coming).
  • killWorkersAfterJob: Kills all the idle workers(no longer required as tasks are completed) except one worker.
Config Keys default accepts
url Undefined String(url)
maxWorkers 3 Number
minWorkers 1 Number
killWorkersAfterJob true Boolean

Note: Url of the worker script is mandatory.

Worker Thread

// Include the library
importScripts('path/to/super-workers.min.js');
// It will expose `SuperWorkers`
var child = new SuperWorkers.WorkerThread();

Constructor doesn't need any configuration but would accept a configurable Object with different keys that would not be used by the library.

API

Refer above section on how to create an instance before hitting API methods.

Main Thread Methods
  • exec

    Executes a task

    Parameter Description Accepts
    method Function that needs to be executed Function(Offloading) / String(Onloading)
    params List of arguments that the function accepts Array
    config Task config Object
      function add (a, b, timeout) {
        var p = SuperWorkers.Promise.defer();
    
        setTimeout(function () {
            return p.resolve(a + b);
        }, timeout || 0);
    
        return p.promise;
      }
    
      var taskConfig = {
        name: 'Add',
        priority: 'LOW',
      };
      sp.exec(add, [1, 2, 2000], taskConfig); // Offloading the method
    

    Define add mthod in worker script

      importScripts('path/to/super-workers.min.js');
    
      function add (a, b) {
        return a + b;
      }
    

    In Main thread,

      var taskConfig = {
        name: 'Add',
        priority: 'HIGH',
      };
      sp.exec('add', [1, 2, 2000], taskConfig); // Onloading the method
    
  • getAllWorkers

    Returns the list of all the workers.

      sp.getAllWorkers();
    
  • getActiveWorkers

    Returns the list of all active workers.

      sp.getActiveWorkers();
    
  • getIdleWorkers

    Returns the list of all idle workers.

      sp.getIdleWorkers();
    
  • getTerminatedWorkers

    Returns the list of all the terminated workers.

      sp.getTerminatedWorkers();
    
  • terminateAllWorkers

    Termiates all the workers irrespective of the status.

      sp.terminateAllWorkers();
    
  • terminateWorker

    Terminated a particular worker by passing the worker-id.

    Parameter Description
    id id of the worker to be terminated
      sp.terminateWorker('57cd47da-d98e-4a2d-814c-9b07cb51059c');
    
  • broadCastAll

    Sends a same message to all the active workers (Message would be sent via postMessage API)

    Parameter Description
    msg msg to be sent
      sp.broadCastAll('Hello my dear Child! A greeting from Parent.'});
    
  • broadCastTo

    Sends a message to a particular active worker (Message would be sent via postMessage API)

    Parameter Description
    id id of the worker to send a msg
    msg msg to be sent
      sp.broadCastTo('57cd47da-d98e-4a2d-814c-9b07cb51059c', 'Hey! Can you run the script: worker.js? Thanks!');
    
Child(Worker) Thread Methods
  • sendMessage

    Sends a message to Main thread.

    Parameter Description Accepts
    msg msg to be sent Object
    origin origin String
      child.sendMessage(obj, `'*'`);
    
  • exposeMethods

    It works as a proxy. Methods which needs to be onloaded should be exposed before they can used in main thread.

    Parameter Description Accepts
    methods list of all functions Object
      child.exposeMethods({
        add: function (a, b) {
          return a + b;
        }
      });
    
Task Queue Methods
  • get

    Returns the task.

    Parameter Description
    id id of the task
      sp.taskQueue.get('34cd47da-d98e-4a2d-814c-9b07cb510578');
    
  • getAll

    Returns the list of all all tasks.

      sp.taskQueue.getAll();
    
  • getActive

    Returns the list of all active tasks.

      sp.taskQueue.getActive();
    
  • getCompleted

    Returns the list of all completed tasks.

      sp.taskQueue.getCompleted();
    
  • getFailed

    Returns the list of all the failed tasks.

      sp.taskQueue.getFailed();
    
  • tasks and allTasks

    Returns tasks which are in queue.

      sp.taskQueue.tasks
    

    Returns all the tasks i.e. pending, queued, active and completed/failed.

      sp.taskQueue.allTasks
    

Browser Support

Tested in Chrome, Firefox and IE(versions >= 9).

Sites using super-workers

Development Stack

  • Webpack based src compilation & bundling and dist generation.
  • ES6 as a source of writing code.
  • Exports in a UMD format so the library works everywhere.
  • Linting with ESLint.
  • ES6 test setup with Karma, Jasmine and isparta.
  • Test coverage with Istanbul and Coveralls.

Process

ES6 source files
       |
       |
    webpack
       |
       +--- babel, eslint
       |
       o--- tests and covergae
       |
  ready to use
     library
  in UMD format

Scripts

  • npm run build - produces production version(minified) of the library under the dist folder
  • npm run dev - produces development version(un minified) of the library and runs a watcher to detect file changes.
  • npm run test - well ... it runs the tests :)

Contributing

CONTRIBUTING.md

Note: If adding site to the list of sites using super-workers, please mention where to verify this in the PR description.

Authors

Varun Malhotra @softvar

Feel like helping?

Please don't hesitate to buy me a cup of coffee :)

paypal

Copyright and license

The MIT license (MIT)

Copyright (c) 2017 Varun Malhotra

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.