Jayadev Misra

Jayadev Misra is a professor and holder of the Schlumberger Centennial chair in Computer Sciences at the University of Texas at Austin. He received his Ph.D. in 1972 from the Johns Hopkins University. He has been a faculty member at the University of Texas at Austin since 1974, except for a sabbatical during 1983-1984 spent at Stanford University. His research interests are in the area of concurrent programming, with emphasis on rigorous methods to improve the programming process. He has been the past editor of several journals in this area, including: Computing Surveys, Journal of the ACM, Information Processing Letters and the Formal Aspects of Computing. He is the author of two books Parallel Program Design: A Foundation, Addison-Wesley, 1988, co-authored with Mani Chandy, and A Discipline of Multiprogramming, Springer-Verlag, 2001. Misra is a fellow of ACM and IEEE; he held the Guggenheim fellowship during 1988-1989. He was the Strachey lecturer at Oxford University in 1996, and he held the Belgian FNRS International Chair of Computer Science in 1990.

Title of the talk at Coordination'05 : A programming model for wide-area computing

Abstract:

The computational pattern inherent in many wide-area applications is this: acquire data from one or more remote services, calculate with these data, and invoke yet other remote services with the results. Additionally, it is often required to invoke alternate services for the same computation to guard against service failure. It should be possible to repeatedly poll a service until it supplies results which meet certain desired criteria, or to ask a service to notify the user when it acquires the appropriate data. And it should be possible to download an application and invoke it locally, or have a service provide the results directly to another service on behalf of the user.

We introduce site as a general term for a basic service. A web service is a site. More generally, a distributed transaction, which can be regarded as an atomic step of a larger computation, is a site. We call the smooth integration of sites orchestration, and Orc is our theory of orchestration of sites. Orchestration requires a better understanding of the kinds of computations that can be performed efficiently over a wide-area network, where the delays associated with communication, unreliability and unavailability of servers, and competition for resources from multiple clients are dominant concerns.

Toby Lehman

Tobin J. Lehman (Toby) joined the IBM Almaden Research Center in 1986, shortly after finishing his Ph.D. degree from the University of Wisconsin-Madison in memory-resident database systems. Toby pioneered many of the techniques used in memory-resident database systems today, such as T-Trees indexes, memory-resident join indexes and dynamically changing granularity concurrency control. After joining IBM, Toby contributed to a number of system projects. He was one of the architects of the first general-purpose server-based backup systems, which was a precursor to today's Tivoli Storage Manager. He created the DB2 Large Object functionality and put the Large Object Types (BLOBs, CLOBs, and DBCLOBs) into the SQL 2 standard. He was the creator of the TSpaces project and remains its chief proponent. He's used TSpaces in a number of coordination roles, such as the foundation piece of a universal print solution, a universal information appliance and a universal multi-sync server. He was also co-founder of the OptimalGrid project, which used TSpaces as a foundation on which to build an autonomic grid system that could solve large, complex, connected, parallel problems on an Internet compute grid.

Title of the talk at Coordination'05 : Whither Coordination Systems: Lessons Learned from the TSpaces Project

Abstract:

When TSpaces was released to the public in 1998, we were excited with all of the possible directions it could take. This model of interaction, the Tuplespace model, featuring asynchronous and anonymous communication, was a natural fit for distributed computing. Coupled with the Java Programming Language, TSpaces could run anywhere and link together just about anything together: any device, any program, or any service. Also, Pervasive Computing was just taking off, and the explosion of portable devices (e.g. PDAs, sensors, smart phones, smart pagers) combined with the development of software services, made it seem that coordination middleware was going to be the new golden child of computing. Thrilled with this prospect, we built or designed many different types of prototypes to demonstrate the power of a coordination middleware system. Eight years, and many Tuplespace-oriented products, projects, or prototypes later, we've validated the underlying model. However, Tuplespace systems, though known and used, are not ubiquitous. They are not included in every copy of MS Windows or Linux (even though there are many who feel they should be). What happened?

In this talk I will give some of the history of the TSpaces project at the IBM Almaden Research Center. I'll describe some of the systems we built on top of TSpaces and discuss their interesting aspects. Then I'll switch gears and talk about the successes and failures of Tuplespace systems, and where I believe the future lies for this under-rated field.