Light-weight hierarchical clustering middleware for public-resource computing
The goal of this work was to investigate ways to implement and improve a public-resource computing middleware. Specifically, to make hosting a public-resource computing project logistically simpler and to examine the affect of hierarchical clustering on bandwidth utilization at the central server. To this end, we present the architecture for our cross-platform, multithreaded public-resource computing middleware.
Implementing and debugging the middleware proved far more challenging than initially anticipated. As hard as debugging multithreaded programs is, our experience has shown us that it can be leveraged to simplify system components. Our main contribution is the final system architecture.