GREE-SC Project Flow Engineering: Using OpenFlow with QoS for Network Traffic Fatma Alali Sujoy Saha Mehdi Mohammadi Yimeng Zhao May 30, 2015.

Slides:



Advertisements
Similar presentations
Diagnosing Wireless TCP Performance Problems: A Case Study Tianbo Kuang, Fang Xiao, and Carey Williamson University of Calgary.
Advertisements

1 Controlling High Bandwidth Aggregates in the Network.
In-Band Flow Establishment for End-to-End QoS in RDRN Saravanan Radhakrishnan.
Optical Ring Networks Research over MAC protocols for optical ring networks with packet switching. MAC protocols divide the ring bandwidth according to.
Traffic shaping with OVS and SDN Ramiro Voicu Caltech LHCOPN/LHCONE, Berkeley, June
Practical TDMA for Datacenter Ethernet
Managing Client Bandwidth in the Presence of Both Real-Time and non Real-Time Network Traffic Maarten Wijnants Wim Lamotte.
Analysis of QoS Arjuna Mithra Sreenivasan. Objectives Explain the different queuing techniques. Describe factors affecting network voice quality. Analyse.
QoS provisioning for large, high-rate file transfers Zhenzhen Yan, Malathi Veeraraghavan, Chris Tracy, and Chin Guok University of Virginia and ESnet Apr.
Network Simulator 2. Introduction Open source network simulator NS uses two languages: C++ and OTcl  C++ is fast to run but slower to change Kernel 
1 Analysis of a window-based flow control mechanism based on TCP Vegas in heterogeneous network environment Hiroyuki Ohsaki Cybermedia Center, Osaka University,
1 Traffic Engineering of High-Rate Large-sized Flows Acknowledgment: UVA work is supported by DOE ASCR grants DE-SC and DE-SC , and NSF grants,
1 Network Simulation and Testing Polly Huang EE NTU
Chapter 30 Quality of Service Copyright © The McGraw-Hill Companies, Inc. Permission required for reproduction or display.
Agenda Background Project goals Project description –General –Implementation –Algorithms Simulation results –Charts –Conclusions.
Ethernet Packet Filtering – Part 2 Øyvind Holmeide 10/28/2014 by.
Troubleshooting Ben Fineman,
Md Baitul Al Sadi, Isaac J. Cushman, Lei Chen, Rami J. Haddad
Chapter 20: Multimedia Systems
High Speed Optical Interconnect Project May08-06
Accelerating Peer-to-Peer Networks for Video Streaming
姓名:劉柏廷 系所:國立中正大學 電機所網路通訊組
Past research work and research work in progress on elephant flows
Low-Latency Software Rate Limiters for Cloud Networks
Instructor Materials Chapter 6: Quality of Service
Chapter 9 Optimizing Network Performance
High Speed Optical Interconnect Project May08-06
Software defined networking: Experimental research on QoS
Integration of and Third-Generation Wireless Data Networks
Chapter 20: Multimedia Systems
QoS & Queuing Theory CS352.
Topics discussed in this section:
Neha Jain Shashwat Yadav
Network Capture and DCF QoS
Mobicom ‘99 Per Johansson, Tony Larsson, Nicklas Hedman
Author: Ragalatha P, Manoj Challa, Sundeep Kumar. K
About CCNP Routing & Switching
SDN API for Access Network Virtualization [1]
Author: Daniel Guija Alcaraz
Transport Protocols over Circuits/VCs
Introduction to Networking
Queue Dynamics with Window Flow Control
Chapter 25 Multimedia TCP/IP Protocol Suite
Chapter 20: Multimedia Systems
© 2008 Cisco Systems, Inc. All rights reserved.Cisco ConfidentialPresentation_ID 1 Chapter 6: Quality of Service Connecting Networks.
Cisco Real Exam Dumps IT-Dumps
Computer Network Performance Measures
School of Electrical Engineering and Computer Science
Congestion Control in Software Define Data Center Network
Group 6-SDN Based Prioritized Information Dissemination
Automatic TCP Buffer Tuning
Adaptive Flow Control using OpenFlow
VoIP Final Report Proposal Spring 2004
A Cognitive Approach for Cross-Layer Performance Management
Provision of Multimedia Services in based Networks
Simulation Results for QoS, pDCF, VDCF, Backoff/Retry
SPEAKER: Yu-Shan Chou ADVISOR: DR. Kai-Wei Ke
Lottery Meets Wireless
OPNET Simulation of Different Queuing Mechanisms
Computer Network Performance Measures
Congestion Control in SDN-Enabled Networks
Iroko A Data Center Emulator for Reinforcement Learning
Packet Scheduling in Linux
Congestion Control Reasons:
Muhammad Niswar Graduate School of Information Science
Introduction to Packet Scheduling
Chapter 20: Multimedia Systems
Congestion Control in SDN-Enabled Networks
Introduction to Packet Scheduling
Wireless MAC Multimedia Extensions Albert Banchs, Witold Pokorski
Presentation transcript:

GREE-SC Project Flow Engineering: Using OpenFlow with QoS for Network Traffic Fatma Alali Sujoy Saha Mehdi Mohammadi Yimeng Zhao May 30, 2015

Content Introduction Resources and Experiment Setup Experiment Results Problem Statement Proposed Approach Experiment Scenario Resources and Experiment Setup Experiment Results Live Demo /17

Problem Statement Mix of large-sized high-rate (elephant) flows and delay-sensitive flows in real networks Alpha flows have adverse affect on the delay-sensitive flows How can we prevent bad effect of Alpha flows on other flows while maintaining alpha flows in their high throughput. /17

Proposed Approach Assumption: The sources of different flows are identified i.e. some hosts produces alpha flow and some others generate non-alpha flow. Solution: Create two separate queues for each flow with a fair scheduling mechanism Alpha flows will be moved to their own separate queue in the switch so that delay-sensitive flows will not be affected /17

Experiment Design To host 4 Ryu High bandwidth for elephant flows and least jitter for delay-sensitive flows queue0 queue1 eth3 OVS eth0 eth1 eth2 eth3 UDP ping nuttcp Host 1 eth1 Host 2 eth1 Host 3 eth1 eth1 Host 4 /17

Experiment Design /17

Used Resources and Tools Kentucky PortoGENI OpenVSwitch Ryu Controller Tools: Linux tools and commands including ssh, tc, curl, nuttcp,… /17

Experiment Set-up All the traffic will be directed to Host 4 Create two queues at eth3 of the OVS node (eth3 is where host 4 is connected) Host 1: Run ping test for 200 s Host 2: Run UDP test (300Mbps) at time 50 and let it run for 150s Host 3: Run nuttcp TCP test at time 100 and let it run for 100 s 200 50 100 Ping H1->H 4 UDP H1->H 4 Nuttcp H1->H 4 8 /17

Experiment Set-up The NIC rate is 1G 1 queue with FIFO scheduler 2 queues with different schedulers and rate: HTB: (70%,30%), (60%,40%), (50%,50%) HTB + SFQ: (70%,30%), (60%,40%), (50%,50%) /17

Experiment Results /17

Experiment Results /17

Experiment Results /17

bandwidth percentage to elephant flow (elephant, other) Experiment Results Measurment bandwidth percentage to elephant flow (elephant, other) (70%,30%) (60%,40%) (50%,50%) HTB_mean (ms) 15.2934 2.6587 2.1428 HTB_std 18.7336 2.5702 1.9665 HTB+SFQ_mean (ms) 1.566 1.6524 1.6457 HTB+SFQ_std 1.3698 1.502 1.4902 /17

Conclusion Our experiment confirms that Alpha flows increase the jitter of delay-sensitive flows With two queues with HTB+SFQ combination we can get good throughput for elephant flows with less jitter for delay- sensitive flows. /17

Future Work Tuning the queuing method to find the optimum configuration. Dynamic identifying of elephant flows and handling with desired QoS support Investigate the effect of different scheduling algorithms on the throughput of the network /17

References Yan, Zhenzhen, Malathi Veeraraghavan, Chris Tracy, and Chin Guok. "On how to provision Quality of Service (QoS) for large dataset transfers." In CTRQ 2013, The Sixth International Conference on Communication Theory, Reliability, and Quality of Service, pp. 13-21. 2013. /17

Live Demo To be presented /17