Self-healing: science, engineering, and fiction

  • Authors:
  • Michael E. Locasto

  • Affiliations:
  • Columbia University

  • Venue:
  • NSPW '07 Proceedings of the 2007 Workshop on New Security Paradigms
  • Year:
  • 2008

Quantified Score

Hi-index 0.00

Visualization

Abstract

Most attacks on computing systems occur rapidly enough to frustrate manual defense or repair. It appears, therefore, that defense systems must include some degree of autonomy. Recent advances have led to an emerging interest in self-healing software as a solution to this problem. It is not clear, however, if the effort to create self-healing mechanisms is actually worth the cost in terms of development effort, deployment complexity, or runtime supervision and monitoring. Furthermore, no general purpose self-healing mechanisms have been shown to be achievable for general systems; it is hard to know beforehand exactly what an application should do in response to an arbitrary vulnerability. A number of very hard problems remain for researchers to explore before self-healing can be reliably applied to real computing systems. This paper provides a critique of the current state of the art and offers the position that self-healing as a concept should be relegated to the status of autonomic computing: a goal worth aiming for as a way to push the boundary of the possible rather than an achievable end in itself. Along the way, we identify a number of important but unsolved research problems in this space.