A MapReduce framework written in Nodejs that coordinates a master node and multiple distributed worker nodes to collectively run map and reduce jobs in parallel. Communication between nodes uses Google's RPC framework gRPC This project is a work in progress.

Currently this only supports WordCount (counting number of occurrences of each word) of the input textfile that is passed to master. I'm working on extending this be able to use any user-defined map or reduce functions.

Installing dependencies (only needed when running locally):

$ npm i

Make sure to copy the text file to be processed into the root directory. This will later be copied into the docker image for master.

Building docker images mr-master and mr-worker:

$ docker-compose build

Running docker-compose to start the containers and begin running MapReduce:

$ docker-compose up

Alternatively, you could also run master locally:

$ node index.js master -f input.txt -m 2 -r 2 -a localhost:5050

and running a worker locally in a different process:

$ node index.js worker -m localhost:5050 -a localhost:6000