Self-organization algorithms for autonomic systems in the SelfLet approach

The difficulties in dealing with increasingly complex information systems that operate in dynamic operational environments ask for self-management policies able to deal intelligently and autonomously with problems and tasks. Biology has been a key source of inspiration in the definition of self-management approaches in the area of computing systems. In this paper we show how some biologically inspired self-organization algorithms have been incorporated into a framework that supports development of autonomic components called SelfLets. The features of a SelfLet include the ability to dynamically change and adapt its internal behaviour according to modifications in the environment, to interact with other SelfLets, in order to provide high-level services, and to make use of autonomic reasoning in order to enable self-* capabilities. In this context, self-organization features represent one of the SelfLets autonomic abilities, and allow them to create groups of SelfLets individuals able to cooperate between each other. The work is complemented with a performance study whose goal is to give insights about strengths and weaknesses of these algorithms.