An efficient energy aware virtual network migration based on genetic algorithm
Frontiers of Computer Science(2019)
摘要
The network virtualization has attached lots of attention from researchers. Most of the prior studies aim at designing embedding solutions for the VN requests to maximize the revenue for infrastructure providers (InPs) while rarely focusing on the energy costs, which is a vital component [1]. Several proposed energy saving methods [2, 3] focus on the energy optimization when the VN requests arrive. However, since the VN requests arrive and depart dynamically and therefore the resource of physical network changes, this optimization needs to be re-optimized. Figure 1 illustrates this context. To tackle the above issue, we proposed an energy aware VN migration based genetic algorithm called EA-VNM-GA. Its core idea is that, since the energy consumption is a linear model to CPU usage with a large offset, if we migrate the virtual nodes from one physical node with light load to other physical node, we can power off that physical node so that the energy can be saved. However, it brings some costs because service interruption for the application running on the virtual nodes occurs during the process of migration. We first model the performance metrics for migration, then we design the EA-VNM-GA algorithm in migration context. In the link re-building stage, we will find physical links with minimum active state physical nodes between the target physical node and the other corresponding physical node for energy conservation.
更多查看译文
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络