Constructing Service Function Chain Test Database: An Optimal Modeling Approach for Coordinated Resource Allocation

Abstract

Network function virtualization (NFV) has become an emerging issue in both academia and industry. By outsourcing network functions from dedicated hardware to virtualization platform, NFV promises to significantly improve the scalability and flexibility of network management and orchestration. One of the main challenges for NFV deployment is to realize coordinated service function chaining on NFV-based infrastructures. This challenge is referred to as the coordinated NFV resource allocation (coordinated NFV-RA) problem which is proved to be NP-hard. In order to response timely to the service variation, many heuristic or meta-heuristic algorithms are proposed to reduce the computing complexity. However, it is very difficult to evaluate the approach degree between obtained sub-optimal solutions and the optima, since finding the optimal solution is a non-trivial task. In this paper, a novel modeling approach called homogeneous link mapping is proposed to find the optimal solutions of a typical three-stage coordinated NFV-RA model with CPLEX. Then we further establish a service function chain (SFC) deployment database with optimal solutions and the results in the database can be used as a criterion to evaluate other SFC algorithms. In order to imitate different practical networks, the SFC deployments are conducted on three type network topologies. And we also analyze the SFC deploying performance on different topologies. At Last, we make the optimal modeling approach open source, and upload the database on http://www.opensource5g.org/database .