Call for Participation

Reliability Analysis of System Failure Data

March 1-2 2007, Cambridge, UK
www.deeds.informatik.tu-darmstadt.de/RAF07
Organized by Microsoft Corporation and TU Darmstadt, Germany

Background:

In relatively short timeframe software controlled systems have progressed from having very bounded functionality and application to now controlling most aspects of our lives. Derivations of the same operating systems can control everything from mobile phones to warships. This explosion in both operational functionality and usage has resulted in the changing context and increasing complexity of the failure modes of these systems. While sophisticated advances have occurred in the collection and analysis of failure data, the academic and industry approaches often cover disparate issues. Industry benefits from having actual data to analyze but their Root Cause Analysis processes are often driven by failure frequency and speed of resolution rather than technical elegance. Academia has developed a plethora of sophisticated techniques but the lack of real data often leaves a gap between the solution and their realistic application by industry. We can continue to bemoan these differences or attempt to bridge this gap between approaches in failure data analysis. This workshop aims for the latter!

Objectives:

The specific intent of this workshop is to bridge this gap by cross-pollination across a joint industry-academic audience through investigating methods to make available real-failure data from a variety of sources (such as Microsoft, IBM, Cisco, etc). People from industry will participate to share experiences of analyzing field failure data. The meeting scope:

• Document experience: state of art & practice on

  • what data to collect - domain sensitivities, accessibility, privacy, utility

  • how to collect data - tools, techniques

  • how to interpret & analyze data?

• What is the current set of complex problems impacting industry?

• New failure analysis (& prediction) techniques, models, tools offering promise?

• Wild thoughts! What should we be doing…what do we do right & wrong?

This first meeting aims at synthesizing the presentations/discussions into a set of tangible and sustained follow up activities including:

• Setting up a data repository (sanitized failure data, tool/techniques repository).

• Identify a bounded set of problems to address through this effort.

• Identify opportunities/processes for joint industry/ academic research.

As a background, refer to the article “Closing the Gap in Failure Analysis”, B. Murphy, M. Garzia, N. Suri; Proc. of DSN-WASR, pp. 59-61, 2006
www.deeds.informatik.tu-darmstadt.de/research

Participation Logistics/Dates:

Participation:

Attendance is strictly by invitation based on the selection of a 2 page (standard IEEE camera ready double column format) position paper that documents (a) new issues/concepts, (b) new academic/ industrial techniques, conceptual/ applied, processes etc, (c) new tools to either obtain or analyze failure data. The selection will be conducted by the organizers based on the technical and contributory match to the meeting objectives.

Registration:

There is no registration fee. All meals will be covered by Microsoft. The participants will need to cover their own travel and accommodation costs at Cambridge.