Hydra

Hydra is a distributed computing framework for Node.js. It provides a way to distribute calculation of some arbitrary function f(x) over arbitrary space S between many computation nodes (browsers or other Node.js processes) connected to master node via websockets.

Install

$ yarn add @onwd/hydra

Usage

Master

1. Create a task:

// master/tasks/example-task.ts

import { Space, Task } from '@onwd/hydra';

function f(x: number): boolean {
  return x * x === 64;
}

const space = new Space({
  values: [0, 1, 2, 3, 4, 5, 6, 7, 8, 9]
});

export const task = new Task({ f, space });

2. Create Master instance and start the server.

// master/index.ts

import { Master } from '@onwd/hydra';
import { task } from './tasks/example-task';

const master = new Master({ task });

master.start();

Worker

Create Worker instance and connect to Master.

import { NodeWebsocketTransport, Worker } from '@onwd/hydra';

const worker = new Worker({
  transport: new NodeWebsocketTransport({
    url: 'ws://localhost:9000'
  })
});

worker.start();

Concept

Hydra provides the following entities:

• Master:
  • stores Task definition as well as algorithm to generate Space and sub-Spaces;
  • handles incoming connections from Workers and stores their current work state;
  • provides the Task definition to Workers as well as sub-Spaces they should work on;
  • stores the Task completion state.
• Worker:
  • connects to Master via Transport and requests a Task definition and a sub-Space to work on;
  • works on a Task using provided definition and a sub-Space;
  • responds to Master with calculation result.
• Transport:
  • allows to use BrowserWebsocketTransport or NodeWebsocketTransport depending on the environment.
• Task:
  • stores f(x) and a Space.
• Space:
  • stores type of the Space, values array or a value generator function.