Second International Workshop on Accelerating Data Management Systems (ADMS) during the 37th International Conference on Very Large Data Bases (VLDB), Seattle, WA, September 2011.
Today, an SSD (Solid State Drive) is essentially a block device attached to a legacy host interface. As a result, the system I/O bus remains a bottleneck, and the abundant flash memory bandwidth as well as the computing capabilities inside SSD are largely untapped. In this paper, we motivate an efficient in-storage computing approach where (part of) data-intensive processing is moved from the host CPU to inside flash SSDs, close to the data source itself ("in-storage processing"). Especially, we focus on accelerating a key database operation, scan. To realize the idea in a cost-effective manner, we deploy special-purpose computing modules using the System-on-Chip technology. While data from flash memory are transferred, a target database operation is applied to the data stream on the fly without any delay. This reduces the amount of data to transfer to the host drastically, and in turn, ensures all components along the data path in an SSD are utilized in a balanced way. Our experimental results show that in-storage processing outperforms conventional CPU based processing by over 13 times for scan operation. It also turns out that in-storage processing can offer sizable energy savings of up to 7x.