The market environment for database machines and servers (keynote address)

Abstract

Since the early 70s, the industrial research community has pursued the elusive dream of a commercially successful database machine: a dedicated-function computer based on an architecture specialized and optimized for database functions, with price and performance characteristics substantially beyond what can be achieved with general purpose software and hardware.During this period, several forces have conspired to frustrate achievement of this goal - forces which are for the most part independent of the DBM research itself. Chief among these is the accelerating pace of advances in microelectronics, which simultaneously creates a moving target for database machine vendors while focusing the beleaguered computer manufacturer's R&D resources on trying to keep up with protracted product life cycles. Meanwhile, successive generations of relational database software products are incorporating sophisticated performance techniques that further challenge the database hardware vendor.The database server, in contrast, finds itself in a far more hospitable environment. At one time, the database machine (say, in the role of a “backend”) and the database server in a network were viewed as minor variations on a common theme. Now the differences are understood to be essential, bringing the server concept in tune with prevailing trends as surely as the backend is in conflict with them. The opportunity for database servers is fueled by the growth of distributed computing and the strength of the “open systems” movement, leading to standards at multiple levels of the relational database architecture.As database server interface standards (de facto or otherwise) are established, a market for these subsystems will emerge which is both very large and broadly based. But, if the market for database machines is to expand beyond narrowly defined niches, product suppliers must overcome far greater obstacles.