Addressing the complexity and scalability of autonomic

点击翻译按钮获取中文摘要

As the Internet grows in scale, it faces an increasingly adverse environment that the original design did not envision. Faults have become the norm rather than the exception, due to component failures, human operational errors, spreads of software viruses, and malicious attacks. In this environment, automated network management techniques, including autonomic, self-configuring and self-optimizing devices and systems as well as self-protecting and self-healing architectures will become increasingly important.The research will focus on the developing the architecture of the configuration module for autonomic networks. The configuration module, - the core capability of an autonomic network -, will be designed to translate the commands and the files or records into abstract structures, does the configuration, adds or activates services according to policy-based goals using the rules of composition provided by the abstract structures and reconvert the configuration result into commands and configuration files. These features are not available today and are merely implemented in form of text-only commands and files. The configuration information including services, policies, functional and version features will be represented by structured tree-like topologies called "constructs" that can be manipulated using a specific algebra with formal composition operations.Our methodology consists first in extending the mathematical model based on configuration logic and formal orchestration constraints enabling the verification of the configuration information through the manipulation of the algebra. We will also use performance simulation to analyze different self-configuration scenarios to assess the self-optimizing, self-healing and self-protecting capabilities. The performance analysis will determine the impacts for various autonomic configuration scenarios. Finally, we plan to experiment with autonomic network implementations and demonstrate its applicability in future systems.