Automated Network Management

Network Management means network monitoring and network devices control, but current network management system can' t support both of them. In order to realize automated network management and control, control theory and graph theory are imported to network management in this thesis.Some mapping regulations from computer networks to graph are defined in Section 2. According to these definitions, network fault management is transferred to graph research. Diffusion Automated Fault Management Algorithm (DAFMA) is proposed in this part and the correctness and time complexity of it are proved carefully. All operations in DAFMA are based on matrix theory and logic operations, so they can be executed easily and quickly. Four classic kinds of fault phenomenon are listed and DAFMA is executed to show how to identify fault location, judge fault reason automated.The Manager-Agent network management model is similar to closed-loop control model and is transferred to be control model in Section 3. In this chapter, how to accumulate sampling time in network control system is analyzed at first, because every network control system will face this problem. Network control system is divided into two kinds, one is switch devices control, the other is server control. Either of them has their special engineering implementation methods. To show how to do network management with control theory, an example about admission control in a Genetic Algorithm computing server is illustrated step by step and their equations and control affection are listed.For testing all kinds of controllers in study, Network Management Technology Test Bed (NMTTB) should be built. The requirements about NMTTB are analyzed, and according to them, software architecture of NMTTB is designed. Structure of every component in NMTTB is described in detail. This architecture is open, scalability and able to support controllers plug-in-play.