Towards Database / Operating System Co-Design

Trends in multicore processors pose serious structural challenges to system software such as databases and operating systems. In this paper we revisit the decades-old problem of the interaction between a database and an operating system in the new context of the changes imposed on both by multicore architectures. Using existing prototypes of a multicore database and multicore operating system, we explore how they can efficiently interact so that the database can make optimal placement and deployment decisions without knowledge of the machine architecture. On diverse hardware, our results show superior memory and interconnect utilization, and dramatic performance improvements in the presence of competing system tasks.