Deakin University
Browse

File(s) under permanent embargo

Reliability-aware virtual network function placement in carrier networks

journal contribution
posted on 2020-03-15, 00:00 authored by L Fang, X Zhang, Keshav SoodKeshav Sood, Y Wang, S Yu
Network Function Virtualization (NFV) is a promising technology that implements Virtual Network Function (VNF) with software on general servers. Traffic needs to go through a set of ordered VNFs, which is called a Service Function Chain (SFC). Rational deployment of VNFs can reduce costs and increase profits for network operators. However, during the deployment of the VNFs, how to guarantee the reliability of SFC requirements while optimizing network resource cost is still an open problem. To this end, we study the problem of reliability-aware VNF placement in carrier networks. In this paper, we firstly redefine the reliability of SFC, which is the product of the reliability of all nodes and physical links in SFC. On this basis, we propose two reliability protection mechanisms: the All-Nodes Protection Mechanism (ANPM) and the Single-Node Protection Mechanism (SNPM). Following this, for each protection mechanism, we formulate the problem as an Integer Linear Programming (ILP) model. Due to the problem complexity, we propose a heuristic algorithm based on Dynamic Programming and Lagrangian Relaxation for each protection mechanism. With extensive simulations using real world topologies, our results show that compared with the benchmark algorithm and ANPM, SNPM can save up to 33.34% and 26.76% network resource cost on average respectively while guaranteeing the reliability requirement of SFC requests, indicating that SNPM performs better than ANPM and has better application potential in carrier networks.

History

Journal

Journal of network and computer applications

Volume

154

Article number

102536

Pagination

1-15

Location

Amsterdam, The Netherlands

ISSN

1084-8045

eISSN

1095-8592

Language

eng

Publication classification

C1 Refereed article in a scholarly journal

Publisher

Elsevier