Using Signatures of Self-Organisation for Monitoring and Influencing Large Scale Autonomic Systems

Abstract

The complexity of large distributed systems makes the monitoring, necessary for self governance and autonomic functions, very problematic. Systems will be ordered hierarchically, consisting of systems of systems. Furthermore a system may be constructed from systems that are fundamentally different in topology from itself. This means that a monitoring strategy that is successful in one part of the system will not necessarily perform well in another system sector. This work considers a phase transition through four system topologies, observed to occur in the evolution of many self organising systems, starting at a regular circular 2D lattice to a scale-free phase. The best monitoring strategy for a scale-free system is considered with a form of acquaintance monitoring proposed. Thus when a system is observed to be acting in a scale-free manner the most suitable monitoring strategy can be deployed. This paper presents two applications that use the results of this work to monitor and influence distinct layers in network topologies