Cluster-Based Repositories and Analysis

We propose to build an Open Analytical Framework using a technology called Cluster Based Repositories to support collaborative, distributed data collection and analysis. The Open Analytical Framework will be an inexpensive, distributed information analysis platform. It emphasizes collaboration through information liquidity and performance through massive scale and parallelization. It is composed of a large distributed ensemble of cooperating data storage nodes (repositories) and applications layered on the resulting uniform data space. We will address the problems of information object identity, schema definition, distributed indexing and query, caching and consistency, change notification, and intermittently connected (e.g., mobile) applications. Our proposed solutions revolve around the use of simple layered data representations (u-forms) and agents to manage the data (shepherds).