Research On Agent-Based Autonomic Computing

Autonomic computing aims to solve management and cost problems in more and more complex IT environment by enabling IT system to be self-management and self-adaptation, which includes self-configuration, self-healing, self-optimization and self-protection. By using some agent technologies for reference, this thesis focuses on autonomic computing system architecture, self-configuration, self-optimization, policy and software engineering for autonomic computing. The main contributions of this dissertation are as follows:(1) From the point of view of Service-Oriented Architecture, an autonomic computing system model framework is proposed. In this framework, a functional structure, a mental state model based on dynamic description logic and an organizational structure for autonomic units are proposed to make them be self-management, and the entire life cycle for autonomic units is given. Moreover, this framework demonstrates what architecture and components autonomic computing system should have as to be self-management, and what model autonomic computing should support for interactions between autonomic units.(2) A universal concept model for policy is proposed, a general policy specification language GPSL is defined based on this model, and a dynamic description logic based goal planning algorithm is given. The policy concept model divides policy into action policy, goal policy, utility policy and mixed policy, and gives detail formalization and classification for action. How to transform goal to action is key for implementing goal policy, based on the mental state model defined for autonomic unit, we give a static planning algorithm and a dynamic planning algorithm to solve this problem.(3) A service-oriented self-configuration method for autonomic computing is proposed. This method divides self-configuration into structural self-configuration used to determine configuration object and interface self-configuration used to configure interface details. Service matchmaking is a key problem in structural self-configuration, so we define a service description language SDLSIN for autonomic units, divide service matchmaking into semantics-based service matchmaking and syntax-based service matchmaking and give service matchmaking algorithms for each of them.(4) A dynamic contract net protocol is proposed to solve structural self-optimization.