Ph.D. Program in Software Engineering, Institute for Software Research
Carnegie Mellon University
Information Representation in Collections of Autonomic Systems
Modern software systems are compositions of autonomic systems. These collections of autonomic systems have both global level objectives and each individual autonomic manager has a set of local objectives. The problem that develops is how to improve performance of the collection against global goals while respecting the autonomy of local systems. The approach to addressing this issue, referred to as Meta-Management, works to address this problem. One element of this approach is the creation of a reusable framework that facilitates the creation of a meta-manager across a broad diverse subset of collections of autonomic systems. One challenge in the creation of this framework is understanding the requirements for the representation of data specific to the need of a meta-manager. This talk will walk through the commonalities among collections of autonomic systems and how those can be exploited to develop a set of requirements for the representation of information for the management of collections of autonomic systems.