Optimizing storage of objects on mass storage systems with robotic devices

Abstract

Automated robotic devices that mount and dismount tape cartridges and optical disks are an important component of a mass storage system. Optimizing database performance in such environments poses additional challenges as the response time to a query may involve several costly volume mounts and dismounts in addition to seek distances within a volume. In this paper we analyze some optimization problems concerning placement of data on such devices. We present a dynamic programming algorithm for optimal loading of data on a robotic device to minimize expected query response time. The method is general in the sense that it can be tailored to work for different hardware characteristics such as seek and mounting times. A variant of the method is also presented which achieves optimal response times subject to storage utilization constraints.