RASSD: A dynamically reconfigurable active storage device for energy efficient data analytics

We study the performance and energy characteristics of a reconfigurable active solid-state drive (RASSD) consisting of a tightly-coupled FPGA/SSD pair. The FPGA implements a compute node that uses partial dynamic reconfiguration to implement hardware accelerators that process data streaming from the SSD. Using K-means clustering as a representative data analytics workload, we show that a RASSD node is 1.3-15.2 times faster than a commodity, multi-core, computer, and consumes 9.4-201.9 times less energy. This suggests that using RASSD nodes in contemporary data centers can have a significant impact on performance and energy efficiency.