Abstract
In the enterprise network, network functions (NFs) are widely used to provide some services, such as security protection and traffic analysis. To deal with the increasing complexity of hardware-based appliances, Network Function Virtualization (NFV) has been proposed to transform from hardware to software. Normally, traffics are required to go through several Virtual Network Functions (VNFs) in a particular order, called service chain. We need to find a single routing path carrying flow demand for each traffic and place VNFs on the path to satisfy the service chain. When satisfying the service chain for a set of traffic with demand, the total flow size go through a link cannot exceed the link's bandwidth. Because of the resource of network is limited, network operator need to decide whether to accept or reject traffics. To increase the network utilization, our goal is to maximize the total demand size of accepted traffic. However, the length of the single routing path is an important issue when solving the this problem. To address these challenges, we study different cases for different length of path. Based on these cases, we introduce a heuristic algorithm to solve this problem with a suitable path length. Our simulations, comparing our algorithm with three heuristics, show that the proposed algorithm achieve a better performance than other heuristics.