A Model Driven Approach for Self-Healing Computing System

Abstract

Self-healing system attempt to enable computing systems to automatically discover, diagnose, and repair (or at least mitigate) faults without human intervention, thus it is a potential approach to improve the dependability and autonomy of computing system. A model-driven approach for modeling and analyzing the self-healing computing system is proposed in this paper. In our approach, the function part and the self-healing part of the system can be modeled respectively. Further, a consequence-oriented fault model is proposed as the ligaments to keep the associations between the function model and the self-healing model. Based on the fault model, we can construct the self-healing model and further analyze the self-heal ability of the computing system. The approach has the advantages of the two aspects: 1) The functional components of target system and the self-healing components for the target system can be modeled respectively and can be coupled through fault model later, 2) The self-heal ability of the target system can be analyzed at the model level.