Prompt damage identification for system survivability
Yanjun Zuo · International Journal of Information and Computer Security · 2008
Survivable systems must identify and isolate any damage as quickly as possible to avoid infection epidemic and outbreak in case of a malicious attack. Any delay during the fault detection and isolation process may lead to system unavailability and is unacceptable in mission-critical applications. In this paper a model is presented to perform damage assessment, fault identification and advance warning. The objective is to help confine the damage propagation (direct or transitive), while making the system survive ongoing attacks and performing necessary self-healing. Our major contribution is the study of the patterns of interconnection communications among applications and the use of communication graphs in damage identification and containment.