Tutorial on Designing and Evaluating Dependable Systems

http://www.crhc.uiuc.edu/EASY/tutorial.html

Saturday, 5 October 2002, San Jose, California, U.S.A.

Held in conjunction with the Second Workshop on Evaluating and Architecting System dependabilitY (EASY)

at the Tenth International Conference on Architectural Support for Programming Languages and Operating Systems (ASPLOS-X)

Speakers

Tutorial description

Given the advances in computer systems performance over the last 20+ years, many researchers traditionally active in the computer architecture, operating systems and programming language communities are turning their attention to other important topics, such as reliability, availability and performability. A crucial step in this path will be understanding the tremendous body of related research from the dependable systems community.

Toward that end, in conjunction with the 2nd EASY (Evaluating and Architecting System dependabilitY) Workshop, we have invited two well-known researchers from the dependability community to present a tutorial on designing and evaluating dependable systems.

The morning half of the tutorial, presented by Dr. Kalbarcyk, will introduce methods for system design, including a broad spectrum of hardware and software error detection and recovery techniques that can be used to build reliable networked systems. Topics include the interplay between hardware and software techniques, techniques for COTS hardware, integration with operating system and network communication layers, and middleware and application-based approaches.

In the afternoon, Dr. Sanders will present methods for validating high-availability systems. After introducing basic concepts, metrics and simple techniques for first-cut rapid validation, the tutorial will introduce several state-of-the-art approaches for more detailed validation using modeling and simulation. This overview will be accompanied by detailed case studies illustrating the techniques. The discussion will consider the pros and cons of these approaches, and provide references for further reading.