Main memory databases have become increasingly important as a way to provide high-performance OLTP (on-line transaction processing) capability. By restricting data to by main memory resident, rather than storing disk resident data entirely in the main memory cache, it is possible to use new technology that results in much better performance. This new setting calls for a revolution in the access methods, concurrency control, and recovery. It reduces the need for high scalability because peak performance is so much higher. This session will explore main memory technologies, the performance that they produce, the problems encountered, and the prospects going forward.
1