NFVdeep: Adaptive online service function chain deployment with deep reinforcement learning

With the evolution of network function virtualization (NFV), diverse network services can be ?exibly o?ered as service function chains (SFCs) consisted of di?erent virtual network functions (VNFs). However, network state and tra?c typically exhibit unpredictable variations due to stochastically arriving requests with di?erent quality of service (QoS) requirements. Thus, an adaptive online SFC deployment approach is needed to handle the real-time network variations and various service requests. In this paper, we ?rstly introduce a Markov decision process (MDP) model to capture the dynamic network state transitions. In order to jointly minimize the operation cost of NFV providers and maximize the total throughput of requests, we propose NFVdeep, an adaptive, online, deep reinforcement learning approach to automatically deploy SFCs for requests with di?erent QoS requirements. Speci?cally, we use a serialization-and-backtracking method to e?ectively deal with large discrete action space. We also adopt a policy gradient based method to improve the training e?ciency and convergence to optimality. Extensive experimental results demonstrate that NFVdeep converges fast in the training process and responds rapidly to arriving requests especially in large, frequently transferred network state space. Consequently, NFVdeep surpasses the state-of-the-art methods by 32.59% higher accepted throughput and 33.29% lower operation cost on average.