We implemented a cluster for independent distributed problem solving using dropbox and a monitoring system via a website and mobile application (cross-platform)

The cluster is a peer to peer computing mechanism allowing for multiple jobs to be distributed to thousands of your friends. The clustering has load balancing features that disallow for traffic overload and bottle necks. The cluster allows for nodes to dynamically connect and disconnect. A computer can join the cluster and add multiple threads to the cluster. The cluster allows for easy join by delivering the necessary files to the scheduled node and automatically returns results to the job owner's dropbox folder. Cluster is resiliant to faults and data loss--auto correcting features Users can deactivate nodes and cluster will schedule according to the active nodes.

Dropbox does not allow for programatically adding shared folders to an account so we wrote a rubby ducky script that injects java script into the dropbox website and auto accepts the requests.

A website and mobile application that monitor live time status about current jobs. Allows for sign in and sign up. Smooth UI. Adjusts to screen sizes. Shows number of jobs completed over time via a live time graph. Shows dynamically which nodes are connected to the cluster and their statistics as well.

Share this project:

Updates