One Size Fits All: An Idea Whose Time Has Come and Gone

Michael Stonebraker, MIT

For the past couple of decades, the major commercial DBMS vendors have been selling “one size fits all”, i.e. a single system appropriate for all data management tasks. In this talk, we discuss three customized engines, one for data warehousing, one for OLTP and one for scientific data processing. In each case, 1-2 orders of magnitude performance advantage is available from a specialized architecture, and we explore the reasons for these dramatic performance differences. We then speculate on the future of the DBMS market, given this information. In particular, we foresee the end of monolithic DBMS solutions.

Speaker Biography

Computer Science and Artificial Intelligence Laboratory M.I.T