Embedding the Minimum Cost SFC with End-to-end Delay Constraint

TL;DR

This paper proposes a heuristic greedy algorithm for embedding service function chains in networks with end-to-end delay constraints, optimizing for minimum cost while handling NP-hard complexity.

Contribution

It introduces a multilevel greedy algorithm to efficiently embed SFCs considering delay constraints and cost, addressing the NP-hard problem.

Findings

The proposed algorithm effectively handles the NP-hard SFC embedding problem.

Simulation results show the algorithm's efficiency and effectiveness.

The method supports flexible network function virtualization deployment.

Abstract

Many network applications, especially the multimedia applications, often deliver flows with high QoS, like end-to-end delay constraint. Flows of these applications usually need to traverse a series of different network functions orderly before reaching to the host in the customer end, which is called the service function chain (SFC). The emergence of network function virtualization (NFV) increases the deployment flexibility of such network functions. In this paper, we present heuristics to embed the SFC for a given flow considering: i) bounded end-to-end delay along the path, and ii) minimum cost of the SFC embedding, where cost and delay can be independent metrics and be attached to both links and nodes. This problem of embedding SFC is NP-hard, which can be reduced to the Knapsack problem. We then design a greedy algorithm which is applied to a multilevel network. The simulation results demonstrate that the multilevel greedy algorithm can efficiently solve the NP-hard problem.