A case for design methodology research in self-* distributed systems

  • Authors:
  • Indranil Gupta;Steven Ko;Nathanael Thompson;Mahvesh Nagda;Chris Devaraj;Ramsés Morales;Jay A. Patel

  • Affiliations:
  • Dept. of Computer Science, University of Illinois at Urbana-Champaign, Urbana, IL;Dept. of Computer Science, University of Illinois at Urbana-Champaign, Urbana, IL;Dept. of Computer Science, University of Illinois at Urbana-Champaign, Urbana, IL;Dept. of Computer Science, University of Illinois at Urbana-Champaign, Urbana, IL;Dept. of Computer Science, University of Illinois at Urbana-Champaign, Urbana, IL;Dept. of Computer Science, University of Illinois at Urbana-Champaign, Urbana, IL;Dept. of Computer Science, University of Illinois at Urbana-Champaign, Urbana, IL

  • Venue:
  • Self-star Properties in Complex Information Systems
  • Year:
  • 2005

Quantified Score

Hi-index 0.00

Visualization

Abstract

We argue that “design methodology research” for self-* distributed systems needs to be recognized and enriched. Methodologies encourage systematic design of distributed protocols. They augment the creative activity of innovation, rather than stifle it. They enable easy design of, and automatic code generation for, distributed systems with predictable properties. Through a taxonomy, we show that methodology research is growing slowly but steadily. As a case study, we present and discuss a new methodology that concretely captures the design of a large class of peer-to-peer distributed hash tables (p2p DHTs) and DHT-based applications. We use this to show some advantages of methodology research, such as effective exploration of the design space for protocols. We also summarize some of our ongoing work in the direction of developing methodologies for distributed protocols.