Abstract
Summary form only given. Autonomic computing has emerged as a paradigm for self managing IT systems to stem the tide of rapidly increasing administration costs in the face of rising IT system complexity. The development of self managing systems poses special challenges to research and development teams. This paper explores some of the development paradigms that lead to successful software projects in autonomic computing. Examples from IBM's DB2 UDB Autonomic Computing project is used so as to highlight recent successes illustrating software engineering principles, algorithmic/mathematical techniques such as online simulation and close loop adaptive control, and experimental results