Abstract

This paper aims at minimizing the communication cost for collecting flow information in Software Defined Networks (SDN). Since flow-based information collecting method requires too much communication cost, and switch-based method proposed recently cannot benefit from controlling flow routing, jointly optimize flow routing and polling switch selection is proposed to reduce the communication cost. To this end, joint optimization problem is formulated as an Integer Linear Programming (ILP) model firstly. Since the ILP model is intractable in large size network, we also design an optimal algorithm for the multi-rooted tree topology and an efficient heuristic algorithm for general topology. According to extensive simulations, it is found that our method can save up to 55.76% communication cost compared with the state-of-the-art switch-based scheme.

Highlights

  • To improve the network performance, network operators need to conduct many measurement tasks

  • Network operators monitor the performance of network devices such as routers and switches to measure the performance of entire network

  • We propose an algorithm to derive an optimal solution for multi-rooted tree topology which is commonly adopted in Data Center Network (DCN)

Read more

Summary

Introduction

To improve the network performance, network operators need to conduct many measurement tasks. There are two types of measurement methods including active measurement and passive measurement. Network operators inject probe traffics into the network. This method is easy to deploy but it may affect the performance of the network since extra traffics are introduced. Network operators monitor the performance of network devices such as routers and switches to measure the performance of entire network. It has little impact to the network performance. The security concerns cannot be ignored since passive measurement requires full access to those network devices

Objectives
Methods
Results
Conclusion
Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call