Architecture Design of a Data Intensive Satellite Image Processing and Distribution System

High-speed and cost-effective architecture design has become a great challenge for future dataintensive scalable high performance computing (HPC) systems. In this paper, we discuss the significant architecture design challenges of a large scale satellite images processing and distribution system maintained by the U.S. Geological Survey (USGS) Earth Resources Observation and Science Center (EROS). Specifically, we conduct a real world case study on 1) how can changing workload greatly affect existing system architecture; 2) how does EROS´s new system architecture adapt to the significant workload changes; and 3) how to further optimize the performance of the current EROS system.