SESSION A

Friday – 8:30-10:00
Convention Ctr - Exhibit Hall 4B

Delivering on the promise of distributed systems

Dr. Alfred Z. Spector
IBM Research

The study of distributed systems in computer science is now many decades old. While concepts of directories, remote procedure call, distributed load balancing, etc. have been around a long time, it has been very difficult to create a successful general purpose platform that supports real distributed applications. In various generations of technology, we have seen the continual reinvention of RPC (e.g., DCE, DCOM, CORBA, RMI, and IIOP) not to mention all the variations on message passing, transaction processing, security models, etc. that seem necessary for real distributed applications. In each generation, we have improved on the flexibility and dynamic nature of distributed interactions, with an eye to increasing their generality and commercial viability. Yet each generation has also stumbled on the way to achieving a usable, open, heterogeneous, distributed environment. Some systems failed because of incompleteness; some because of complexity; some because of inadequate tools; and others because they were proprietary or could not support heterogeneity.

Finally, with the advent of Web Services from the W3C, the promise of a widely accepted, open, heterogeneous, distributed environment may be at hand. Significantly, Web Services depend on the key proposition taught to us by the object oriented community - that scalability and designability depend on encapsulation (with clean interfaces) to reduce complexity. But we must also learn from the dark side of OO: we already know that good tools, useability, and performance will make or break the success of Web Services. Experiences from the history of both distributed systems and object orientation can teach us quite a lot about how to succeed this time with Web Services. This talk will place Web Services in the context of distributed computing's history and propose a roadmap for how it can succeed as the basis for the open platform for distributed computing in the future.

Speaker

Dr. Alfred Z. Spector is vice president of Services and Software in IBM's Research Division responsible for setting IBM's worldwide research strategy in support of its Services and Software businesses. Recently, Dr. Spector was an Adjunct Professor at Columbia University's Computer Science Department and Senior Technical Strategist in IBM's Application and Integration Middleware (AIM) business, which has responsibility for a number of IBM software product families including CICS, WebSphere, MQSeries, WebSphere Studio and Eclipse. Previously, Dr. Spector was the general manager of Marketing and Strategy for IBM's AIM business, and the general manager of IBM's Transaction Systems business. Dr. Spector was also founder and CEO of Transarc Corporation, a pioneer in distributed transaction processing (Encina, DCE) and wide area file systems (DFS, AFS), and an Associate Professor of Computer Science at Carnegie Mellon University. Prior to joining Carnegie Mellon, Dr. Spector was affiliated with the IBM San Jose (now Almaden) Research Laboratory while in graduate school.

Dr. Spector received his Ph.D. in Computer Science from Stanford University on the topic of multiprocessing architectures for local area computer networks and his A.B. in Applied Mathematics, Magna cum Laude, from Harvard University.