Multiple Service Chain Placement and Routing in a Network-enabled Cloud

TL;DR

This paper introduces a novel column generation model for the complex problem of placing and routing multiple Virtual Network Function Service Chains in network-enabled clouds, aiming to optimize resource use.

Contribution

It presents the first explicit focus on multiple VNF service chain placement and routing, reducing computational complexity with a new model and demonstrating benefits of network-enabled clouds.

Findings

Network-enabled clouds reduce resource consumption compared to centralized data centers.

The proposed model effectively handles complex placement and routing problems.

Distributed NFVI avoids infeasibility issues present in centralized setups.

Abstract

Network Function Virtualization (NFV) aims to abstract the functionality of traditional proprietary hardware into software as Virtual Network Functions (VNFs), which can run on commercial off the shelf (COTS) servers. Besides reducing dependency on proprietary support, NFV helps network operators to deploy multiple services in a agile fashion. Service deployment involves placement and in sequence routing through VNFs comprising a Service Chain (SC). Our study is the first to focus on the computationally complex problem of multiple VNF SC placement and routing while considering VNF service chaining explicitly. We propose a novel column generation model for placing multiple VNF SCs and routing, which reduces the computational complexity of the problem significantly. Our aim here is to determine the ideal NFV Infrastructure (NFVI) for minimizing network resource consumption. Our results indicate that a Network enabled Cloud (NeC) results in lower networkresource consumption than a centralized NFVI (e.g., Data Center) while avoiding infeasibility with a distributed NFVI.