Optimized Virtual Network Functions Migration for NFV
Jing Xia, Zhiping Cai, Ming Xu · 2016
Combining with software-defined networking and IT virtualization technologies, Network Function Virtualization (NFV) has been proposed as an important technology to speed up deployment of new network services. VNF (Virtual Network Function) migration is a critical step to redeploy virtual network functions for providing better network services. Previous work in virtual network function migration primarily focused on the migration mechanism, including maintaining internal state consistency and reducing migrating time. In this paper, we address the problem of optimally migrating virtual network functions. As the computing and network resource requirement of virtual network functions have been changed, these virtual network functions have to be migrated to meet the computing and network resource constraints. As the SDN controllers conducted technology is used to migrate the virtual network functions, the migration cost depends on the buffer size of controllers and the time of transferring the internal state of virtual network functions. A cost model is proposed to evaluate the migration cost. The problem of optimal virtual network functions migration with satisfying computing and network resource constraints is NP-hard. A heuristic algorithm is proposed for computing the approximate solution. The effectiveness of the algorithms is validated by simulations evaluation.