Latency-Aware Virtual Network Function Chain Recovery in Provider Networks with Network Slicing

With the rapid development of network function virtualization (NFV) technology, network services can be provided by virtual network functions (VNFs) deployed in provider networks. One of the most important related issues is to install duplicated VNFs to avoid the failures of VNFs. Due to the fact that duplicated VNFs will occupy lots of resources in networks, some research works study on how to efficiently recover failed VNFs to satisfy the demands of users recently. However, previous studies often focus on recovering demands, without considering if the latency of the recovered demands can be satisfied or not. While considering the latency constraint of demands, in this paper, we study on the problem of recovering failed VNFs to recover failed demands such that the total weight of the recovered demands is maximized, while still ensuring the latency requirement of the recovered demands, where the weight of a demand is used to represent its importance. For this problem, an efficient algorithm, termed the latency-aware recovery algorithm (LARA), is proposed. In addition, simulations are conducted to show the performance of the LARA.