Specification and implementation of autonomic large-scale system behaviors using domain specific modeling language tools

Space applications are often highly complex systems composed of a large number of hardware and software components. The harsh environments in which the hardware operates and long mission lifetime make this class of systems susceptible to component failures, jeopardizing the success of space missions. Embedding autonomic fault-mitigation behaviors in the system can greatly increase the chance of mission success and decrease catastrophic consequences. Work at Vanderbilt University has resulted in a tool for designing fault-adaptive, autonomic systems. The tool provides an environment that supports specification of custom fault-adaptive behaviors and automated generation of behavior code. These behaviors are defined to perform application-specific adaptations in response to external and internal fault events. This paper describes the application of the tool in managing faults on large-scale embedded computing clusters for a high-energy physics instrumentation application and how this tool can be extended to space applications.