A Self-repair Architecture for Cluster Systems

摘要

The paper presents the Jade framework for the construction of self-repairable cluster systems. Jade adopts an architecture-based approach to management, and maintains a causally connected view of the software architecture of the managed system, itself configured and manipulated as a component-based structure. Self-repair is achieved through a combination of component-based design, reflection and active replication of the management subsystem. The paper illustrates the benefits of the Jade approach through its application to a JEE Web application server. Specifically, our evaluation shows that the JADE framework adds negligible overhead to the operation of a managed system, and that Jade achieves short MTTR even with a simple repair policy.