In order to meet the requirements of emerging services, the future Internet will need to be flexible, reactive and adaptive. Network management functionality is essential in providing dynamic reactiveness and adaptability but current network management approaches have limitations and are inadequate to meet the relevant demands. In search for a paradigm shift, recent research efforts have been focusing on self-management principles. The PhD work presented in this paper proposes to investigate how autonomic principles can be extended and applied to fixed networks for quality of service (QoS) and performance management. The paper describes the two main research issues that will be addressed, namely (a) coordinated decision making in distributed environments, and (b) lightweight learning capabilities, and highlights their importance on realistic application scenarios for the emerging Internet.