Fault Recovery Designs for Processor-Embedded Distributed Storage Architectures with I/O-Intensive DB Workloads

Fault recovery has become an essential capability for systems that process large data-intensive workloads. Processor-embedded distributed storage architectures offload user-level processing to the peripheral from the host servers. Our earlier work investigated the performance benefits of such architectures for disk- and MEMS-based smart storage devices. In this paper, we focus on the issue of fault recovery. We propose recovery schemes for TPC-H based workloads, and evaluate several recovery scenarios applicable to both disk- and MEMS-based smart storage architectures.