Dagstuhl Seminar 9436
Unifying Theory and Practice in Distributed Systems
( Sep 05 – Sep 09, 1994 )
Permalink
Organizers
- A. Schiper
- F. Cristian
- F. Mattern
- K. Birman
Contact
During the past 20 years, a substantial theoretical and practical base has evolved in the area of distributed computing. However, this work has been done by largely disjoint sets of researchers, with the result that much theory is inapplicable to real—world systems, and many of the real-world systems that have been built suffer from weaknesses that could be overcome using the existing theoretical methodology. It was therefore the intention of the seminar “Unifying Theory and Practice in Distributed Systems” to bring together a diverse group of pragmatically inclined theoreticians and theoretically inclined practitioners with the goal of sharing insight, educating one another, and laying the groundwork for the next generation of distributed systems research and development.
The Dagstuhl seminar was organized by Kenneth Birman (Cornell University), Flaviu Cristian (University of California at San Diego), Friedemann Mattern (University of Saarland at Saarbrücken), and André Schiper (Ecole Polytechnique de Lausanne). It brought together 38 participants — established experts from academia and industry as well as young scientists — from nine countries. 30 talks (some with on-line and video demonstrations) were given during the week, often followed by lively and sometimes controversial discussions. The emphasis of the talks was on the following general themes:
- important paradigms and influential concepts
- fundamental algorithms and principles
- fault—tolerance
- real-time
- system structures, basic services, toolkits
- large scale issues, application domains, case studies
In two evening discussion sessions the implications of the Fischer-Lynch-Paterson Theorem (impossibility of distributed consensus with one faulty processor) for practical system design and the various notions of “real time” were discussed. In total, the seminar was considered to be successful and very interesting, and the publication of proceedings is now considered.
The special nature and atmosphere of Schloss Dagstuhl offered ample opportunities for personal discussions, the computer science library was also used extensively by some participants. The fine food, the good wine, and the perfect organization by the office and the local staff of Schloss Dagstuhl was much appreciated by all participants. If there is anything to blame, then it is the weather—we hope that the next time we will have some sunny days!
- A. Schiper
- F. Cristian
- F. Mattern
- K. Birman