Flash Device Support for Database Management

While disks have oered a stable behavior for decades -thus guaranteeing the timelessness of many database design deci- sions, ash devices keep on mutating. Their behavior varies across models, across rmware updates and possibly in time for the same model. Many researchers have proposed to adapt database algorithms for existing ash devices; others have tried to capture the performance characteristics of ash devices. However, today, we neither have a reference DBMS design nor a performance model for ash devices: database researchers are running after ash memory technology . In this paper, we take the reverse approach and we dene how ash devices should support database management. We ad- vocate that ash devices should provide DBMS with more control over IO behavior without sacricing correctness or robustness, exposing the full potential of the underlying ash chips in terms of performance. We suggest two approaches: (a) keep the narrow block device interface, or (b) provide a rich interface that allows a DBMS to explicitly control IO behavior. We believe that these approaches are natural evolutions of the current generation of ash devices, whose complexity and opacity is ill-suited for database manage- ment. We describe the design space for the two proposed approaches, discuss how they would benet many existing techniques proposed by the database research community, and identify a set of new research issues.