Service Function Chain Update and Buffer Scheduling Model With State Consistency

A service function chain (SFC) is a set of virtualized network functions (VNFs) concatenated in a virtualized network. Paths of SFCs need to be updated to achieve load balancing and security. When updating SFCs, states of VNF instances need to be migrated to new VNF instances; otherwise, the processing accuracy of packets degrades. While the state migration is performed, packets need to be buffered to guarantee state consistency. In existing models, if the state migration to a VNF instance is not completed, packets are buffered at that VNF instance. When these models are adopted to the update of multiple SFCs, the buffered packets can be concentrated at certain VNF instances and packet loss can occur. This paper proposes an SFC update and buffer scheduling model with the state consistency. In the proposed model, packets can be buffered at destination VNF instances for an arbitrary time. The proposed model is formulated as a mixed integer linear programming (MILP) problem, whose objective is to minimize the maximum amount of buffered packets among VNF instances. The decision problem of the proposed model is proved to be NP-complete. Two algorithms are introduced for the case where the MILP problem is not tractable. Numerical results show that the proposed model can suppress the maximum amount of buffered packets compared to the benchmark model by more than 40% in several scenarios.