| Summary: | 碩士 === 銘傳大學 === 資訊工程學系碩士班 === 99 === Cloud Computing has been a very popular topic recently. One of the applications in the Cloud Computing is virtualization. Network virtualization has emerged as a powerful way to allow multiple virtual networks, each customized to a particular application, to run on a common substrate network. Our research consists of two parts. The first is a node mapping algorithm and a link mapping algorithm. The second is a path migration algorithm. The first part focuses on how to use the proposed node mapping algorithm and link mapping algorithm to map the virtual networks to a substrate network. The second part focuses on how to use the proposed path migration algorithm to migrate virtual links to different substrate paths, which can improve the substrate’s ability to accept more virtual networks.
The study of the first part concentrates on mapping problems. The mapping problems consist of two parts. The first is node mapping algorithm, which focus on how to map the virtual nodes to substrate nodes. The greedy algorithm is proposed to assign the virtual nodes to substrate nodes. The second is link mapping algorithm. Link mapping algorithm also consists of two problems. One is that a virtual link is mapped to a single substrate path. The widest path algorithm and cut-shortest path algorithm are proposed for this problem. The other is that a virtual link is mapped to multiple substrate paths. In this case, path diversity is enabled in the substrate network. Cut-shortest path algorithm proposed for this problem.
The study of the second part focuses on path migration in the substrate network. When a new virtual network requests to map to a substrate network, it is possible that no resource in the substrate network can meet the requirement of the new virtual network. In this situation, path migration must be enabled to re-arrange all the virtual networks that have already been mapped to the substrate network. The proposed path migration algorithm consists of three steps. First, the virtual nodes in the new virtual network are mapped to the substrate nodes by using node mapping algorithm. Second, the algorithm selects one existing virtual network to migrate to other substrate links. Third, if the resource in the substrate can meet the requirement of the new virtual network, the path migration algorithm stops. Otherwise, the algorithm continues to migrate the next existing virtual network. The cut-shortest path algorithm is proposed for path migration.
|
|---|
