Parameter Selection for Autonomic Deployment Decision Making

Today's computing networks are built to have a very high degree of heterogeneity, complexity, and interrelation. To be able to interact efficiently with those networks, scientists and computing system developers equip soft- and hardware with the ability to solve subtasks in an autonomic fashion. Especially with regard to easily accessible, content-rich, and increasingly pervasive networks, the question arises how best to select among multiple alternatives to deploy services or to obtain data in an efficient and preferably autonomic way. In this paper we address this question and present a deployment decision making (DDM) system accordingly. In particular, we concentrate on the performance-optimizing effects of selecting parameters and parameter subsets for the decision making process. These effects are deduced analytically and illustrated by an experimental evaluation.