Research On Virtual Network Mapping Algorithm Based On Load Pressure

With the rapid expansion of network scale and the emerge of new applications, the problems of current Internet architecture have become more and more serious, especially in security, mobility, scalability and energy consumption. To solve these problems,more and more new agreement and contents were added to the network. But making too much cosmetic changes and patching up only make the Internet more and more overstaffing and rigid. As an important means to solve the rigid problem of current Internet, network virtualization technology has attracted much attention of the researchers from all over the world. How to integrate multiple heterogeneous virtual network mapping to the same underlying network, namely virtual network mapping, is a big problem to be solved in the development of the virtual network technology. Therefore, researching on the virtual network mapping algorithm has important theoretical meaning and practical application value.The paper reviews the classification of the virtual network mapping algorithm and analyzes the advantages and disadvantages of all kinds of mapping strategy. Two phases mapping strategy is adopted because it has a good balance in the complexity of algorithm and mapping effects. The problem of node mapping and link mapping under network virtualization environments is NP-hard and scientists solve it with heuristic algorithm at present. The paper makes a series of improvements to the existing virtual network mapping algorithm from the aspects of optimizing underlying network pressure and proposes a new virtual network mapping algorithm.First, in the node mapping phase, the paper proposes a virtual node mapping algorithm that based on pressure equalizing with the goal of optimizing the node pressure. The algorithm is inspired by water injection in wireless management principle and combines the advantages of the greedy algorithm. It will choose the node with the most residual resource to map when the physical nodes have plenty available resources. And when the residual resource of the physical nodes are insufficient, the algorithm will choose the node with the least pressure. By combining the remaining resources and node pressure, this mapping strategy can improve the physical network load balancing performance and reduce the possibility of the bottleneck nodes occur. Then, in the link mapping phase, through improving the traditional multiple commodity flow algorithm, the paper proposes a virtual link mapping algorithm based on the stress state feedback. The algorithm designs a remaining bandwidth feedback factor to effectively distinguish the mapping receptivity of physical links. By combining the virtual link mapping function with the physical link load condition, the physical network can allocates the link resource adaptively, lower the link pressure and improve the load balancing ability. Finally, the paper combines the improved node mapping algorithm and link algorithm, and compares with other mapping scheme by simulation. The simulation results show that virtual network mapping algorithm based on load pressure can reduce the node stress and link stress. It can also improve the load balancing performance of the physical network and improve resource utilization.