Hop-by-hop bandwidth allocation and deployment for SFC with end-to-end delay QoS guarantees

Abstract

Network function virtualization and 5G network slicing technology can provide a more resilient and manageable Internet by deploying various requests on different slices. In this paper, we propose a hop-by-hop bandwidth allocation framework for service function chains (SFCs) to guarantee delay QoS constraints of requests. A end-to-end (E2E) SFC contains multiple virtual network functions (VNFs), which is constructed as a tandem queueing system. The arrival is modeled as an interrupted bernoulli process (IBP) or two-aggregation Markov modulated bernoulli process (2A-MMBP) to represent the traffic of typical services, data or video call. And the effective bandwidth (EB) formulas for these intricate Markovian processes are derived. Then we use EB/ effective capacity (EC) theory to allocate the bandwidth for SFCs hop by hop and deduce the service probabilities of the nodes. On the basis of EB/EC bandwidth allocation, the departure process is fitted into an IBP or a MMBP and then becomes the arrival of the next hop, which is beneficial to engineering implementation. The novel delay evaluation metrics, E2E packet delay and delay violation probability, are developed to assess the performance of SFC deployment. We formulate the SFC deployment as an integral linear programming (ILP) problem with the objective of minimizing the total cost. Finally, a heuristic algorithm called hop-by-hop bandwidth allocation and SFC deployment (HBASD) is designed to solve this problem. Simulation results show that compared with the up-to-date methods, our algorithm can guarantee the QoS requirements of E2E delay for services and save resources.