1. 2019/7/26
OpenFlow-Enabled User Traffic Profiling in Campus Software Defined Networks
Presenter: Wei-Li,Wang
Date: 2016/1/4
Author: Taimur Bakhshi and Bogdan Ghita
2016 IEEE 12th International Conference on Wireless and Mobile Computing, Networking and Communications (WiMob)
Department of Computer Science and Information Engineering
National Cheng Kung University, Taiwan R.O.C.
CSIE CIAL Lab 1

2. 2019/7/26
Introduction
OpenFlow protocol, providing per-flow monitoring and management of OpenFlow compliant SDN switches.
The OpenFlow protocol also caters for improving the individual service performance by guaranteeing quality of service through isolated application flow metering.
National Cheng Kung University CSIE Computer & Internet Architecture Lab
CSIE CIAL Lab

3. Introduction
However, isolated application performance, which is the default traffic optimization method in SDN, may not be suitable for all campus users, particularly the users using different set of applications to the ones optimized.
National Cheng Kung University CSIE Computer & Internet Architecture Lab

4. Introduction
Profiling user traffic based on application trends may more accurately express user activities and aid administrators in aligning optimization solutions to the inherent campus user classes instead of individual applications.
National Cheng Kung University CSIE Computer & Internet Architecture Lab

5. Methodology
The proposed traffic profiling methodology comprises of two main components (i) OpenFlow traffic monitor and (ii) the traffic profiling engine.
National Cheng Kung University CSIE Computer & Internet Architecture Lab

6. Methodology – Traffic Monitor
National Cheng Kung University CSIE Computer & Internet Architecture Lab

7. Methodology – Traffic Monitor
Collected from a realistic academic network.
Two weeks, 42 users.
Linux monitoring machine (VM1) running an Open vSwitch (SW1) and Ryu SDN controller instance connected to the departmental LAN.
Port monitoring was enabled at the default gateway (SW2) to replicate all traffic to and from each user to the VM1 interface (virtual switch SW1).
National Cheng Kung University CSIE Computer & Internet Architecture Lab

8. Methodology – Traffic Monitor
2019/7/26
Methodology – Traffic Monitor
Ryu controller to collect per user flow statistics. All user machines used static IP
National Cheng Kung University CSIE Computer & Internet Architecture Lab
CSIE CIAL Lab

9. Methodology – Traffic Monitor
Flow installation.
packet_in
packet_out
Statistics collection
flow_stats
table_stats
flow-removed
National Cheng Kung University CSIE Computer & Internet Architecture Lab

10. Methodology – Traffic Monitor
2019/7/26
Methodology – Traffic Monitor
Since the purpose of the experiment was data collection and not actual flow forwarding and OpenFlow does not prevent flow installation towards a blocked port, virtual port2 on SW1 was set to blocking mode (sink).
National Cheng Kung University CSIE Computer & Internet Architecture Lab
CSIE CIAL Lab

11. Methodology – Traffic Profiling Engine
User traffic collected by the traffic monitor was classified by matching seven-tuple traffic records against source and destination IP addresses and ports used by the respective users and campus servers.
National Cheng Kung University CSIE Computer & Internet Architecture Lab

12. Methodology – Traffic Profiling Engine
National Cheng Kung University CSIE Computer & Internet Architecture Lab

13. Methodology – Traffic Profiling Engine
National Cheng Kung University CSIE Computer & Internet Architecture Lab

14. User Traffic Profile
As shown in Fig. 5, the variance between individual values is maximum until k=6, however, subsequent values of k (≥6) show minimum change in the successive overall variance (<0.05%). Therefore, for the present study, k=6 provided an optimal number of user profiles fitting the sample space used for further analysis.
National Cheng Kung University CSIE Computer & Internet Architecture Lab

15. User Traffic Profile – Extracted Profiles
2019/7/26
User Traffic Profile – Extracted Profiles
web browsing (39%) Communications(64.5%) Enterprise(50.3)
National Cheng Kung University CSIE Computer & Internet Architecture Lab
CSIE CIAL Lab

16. User Traffic Profile – Extracted Profiles
National Cheng Kung University CSIE Computer & Internet Architecture Lab

17. User Traffic Profile – Extracted Profiles
National Cheng Kung University CSIE Computer & Internet Architecture Lab

18. User Traffic Profile – Extracted Profiles
National Cheng Kung University CSIE Computer & Internet Architecture Lab

19. User Traffic Profile – Profiling Computational Cost
National Cheng Kung University CSIE Computer & Internet Architecture Lab

20. User Traffic Profile – Control Channel Overhead
National Cheng Kung University CSIE Computer & Internet Architecture Lab

21. User Traffic Profile – Control Channel Overhead
National Cheng Kung University CSIE Computer & Internet Architecture Lab

22. User Traffic Profile – Control Channel Overhead
National Cheng Kung University CSIE Computer & Internet Architecture Lab