Distributed Configuration Management for Reconfigurable Cluster Computing

Cluster computing offers many advantages as a highly cost-effective and often scalable approach for highperformance computing in general, and most recently as a basis for hardware-reconfigurable systems. To achieve the full potential of performance of reconfigurable HPC systems, a runtime configuration service is required. Centralized configuration services are a natural starting point but tend to limit performance and scalability. For large-scale RC systems, the configuration service must be optimized for the system topology and management scheme. This paper presents the CARMA framework for reconfigurable clusters including several distributed configuration management schemes that leverage highspeed networking. The experimental results from this paper highlight the affects of the design of this configuration service and provide a comprehensive performance analysis of two of the proposed management schemes.