The goal of this research is to advance the MapReduce framework for large-scale distributed computing across multiple data centers with multiple clusters. The framework supports distributed data-intensive computation among multiple administrative domains using existing unmodified MapReduce applications