Performance Benchmarking of Network Function Chain Placement Algorithms.

The Network Function Virtualization (NFV) paradigm enables new flexibility and possibilities in the deployment and operation of network services. Finding the best arrangement of such service chains poses new optimization problems, comprising a combination of placement and routing decisions. While there are many algorithms on this topic proposed in literature, this work is focused on their evaluation and on the choice of reference for meaningful assessments. Our contribution comprises two problem generation strategies with predefined optima for benchmarking purposes, supplemented by an integer program to obtain optimal solutions in arbitrary graphs, as well as a general overview of concepts and methodology for solving and evaluating problems. In addition, a short evaluation demonstrates their applicability and shows possible directions for future work in this area.