TCP over OBS Joint Work with F. Baccelli, D. Hong, G. Petit, F. Poppe Julien Reynier Article available at

Slides:



Advertisements
Similar presentations
Congestion Control and Fairness Models Nick Feamster CS 4251 Computer Networking II Spring 2008.
Advertisements

QoS Strategy in DiffServ aware MPLS environment Teerapat Sanguankotchakorn, D.Eng. Telecommunications Program, School of Advanced Technologies Asian Institute.
Spring 2000CS 4611 Introduction Outline Statistical Multiplexing Inter-Process Communication Network Architecture Performance Metrics.
Principles of Congestion Control Chapter 3.6 Computer Networking: A top-down approach.
Congestion Control Created by M Bateman, A Ruddle & C Allison As part of the TCP View project.
CS640: Introduction to Computer Networks Mozafar Bag-Mohammadi Lecture 3 TCP Congestion Control.
Computer Networking Lecture 17 – TCP & Congestion Control Dejian Ye, Liu Xin.
Congestion Control: TCP & DC-TCP Swarun Kumar With Slides From: Prof. Katabi, Alizadeh et al.
A performance model for an asynchronous optical buffer W. Rogiest K. Laevens D. Fiems H. Bruneel SMACS Research Group Ghent University Performance 2005.
TCP/IP Performance across Optical Packet-Switched (OPS) Networks Jingyi He ( Co-supervised by Dr. Gary Chan and Dr. Danny Tsang ) Apr. 25, 2002.
XCP: Congestion Control for High Bandwidth-Delay Product Network Dina Katabi, Mark Handley and Charlie Rohrs Presented by Ao-Jan Su.
Chapter 3 Transport Layer slides are modified from J. Kurose & K. Ross CPE 400 / 600 Computer Communication Networks Lecture 12.
Dynamic Internet Congestion with Bursts Stefan Schmid Roger Wattenhofer Distributed Computing Group, ETH Zurich 13th International Conference On High Performance.
High Performance All-Optical Networks with Small Buffers Yashar Ganjali High Performance Networking Group Stanford University
Transport Layer3-1 Congestion Control. Transport Layer3-2 Principles of Congestion Control Congestion: r informally: “too many sources sending too much.
Lecture 2 Introduction 1-1 Chapter 1: roadmap 1.1 What is the Internet? 1.2 Network edge  end systems, access networks, links 1.3 Network core  circuit.
1 Architectural Results in the Optical Router Project Da Chuang, Isaac Keslassy, Nick McKeown High Performance Networking Group
FF-1 9/30/2003 UTD Practical Priority Contention Resolution for Slotted Optical Burst Switching Networks Farid Farahmand The University of Texas at Dallas.
1 Chapter 3 Transport Layer. 2 Chapter 3 outline 3.1 Transport-layer services 3.2 Multiplexing and demultiplexing 3.3 Connectionless transport: UDP 3.4.
Data Communication and Networks
Lecture Internet Overview: roadmap 1.1 What is the Internet? 1.2 Network edge  end systems, access networks, links 1.3 Network core  circuit switching,
On The Deflection Routing in QoS Supported Optical Burst-Switched Network Ching-Fang Hsu, Te-Lung Liu, and Nen-Fu Huang Dep. of CS, NTHU, Taiwan, ROC.
UCB Review for M2 Jean Walrand U.C. Berkeley
Lecture Internet Overview: roadmap 1.1 What is the Internet? 1.2 Network edge  end systems, access networks, links 1.3 Network core  circuit switching,
Jennifer Rexford Fall 2014 (TTh 3:00-4:20 in CS 105) COS 561: Advanced Computer Networks TCP.
Outline Introduction Switching Techniques Optical Burst Switching
References TCP is Max-Plus Linear, [Single flow] F.Baccelli, D.Hong TCP is Max-Plus Linear, Sigcomm 00 AIMD, Fairness and Fractal Scaling of TCP traffic,
Multipath Protocol for Delay-Sensitive Traffic Jennifer Rexford Princeton University Joint work with Umar Javed, Martin Suchara, and Jiayue He
Optical Switching Switch Fabrics, Techniques and Architectures 원종호 (INC lab) Oct 30, 2006.
Lecture 1, 1Spring 2003, COM1337/3501Computer Communication Networks Rajmohan Rajaraman COM1337/3501 Textbook: Computer Networks: A Systems Approach, L.
All Optical Switching Architectures. Introduction Optical switches are necessary for achieving reliable, fast and flexible modular communication means.
3: Transport Layer3b-1 Principles of Congestion Control Congestion: r informally: “too many sources sending too much data too fast for network to handle”
Transport Layer 4 2: Transport Layer 4.
Transport Layer3-1 Chapter 3 outline r 3.1 Transport-layer services r 3.2 Multiplexing and demultiplexing r 3.3 Connectionless transport: UDP r 3.4 Principles.
CS144 An Introduction to Computer Networks
Understanding the Performance of TCP Pacing Amit Aggarwal, Stefan Savage, Thomas Anderson Department of Computer Science and Engineering University of.
Principles of Congestion Control Congestion: informally: “too many sources sending too much data too fast for network to handle” different from flow control!
1 Optical Burst Switching (OBS). 2 Optical Internet IP runs over an all-optical WDM layer –OXCs interconnected by fiber links –IP routers attached to.
TTM1: ”Burst, packet and hybrid switching in the optical core network” Steinar Bjørnstad et al.
IP Traffic Management In IP- OVER-WDM Networks:HOW? Javier Aracil, Daniel Morato, and Mikel Izal Universidad Publica de Navarra, Pamplona, Spain Presenter:Chen.
Contents Causes and cost of congestion Three examples How to handle congestion End-to-end Network-assisted TCP congestion control ATM ABR congestion control.
TCP Trunking: Design, Implementation and Performance H.T. Kung and S. Y. Wang.
1 Optical Packet Switching Techniques Walter Picco MS Thesis Defense December 2001 Fabio Neri, Marco Ajmone Marsan Telecommunication Networks Group
Prentice HallHigh Performance TCP/IP Networking, Hassan-Jain Chapter 8 TCP/IP Performance over Optical Networks.
Analysis of Buffer Size in Core Routers by Arthur Dick Supervisor Anirban Mahanti.
T. S. Eugene Ngeugeneng at cs.rice.edu Rice University1 COMP/ELEC 429/556 Introduction to Computer Networks Principles of Congestion Control Some slides.
By N.Gopinath AP/CSE Unit: III Introduction to Transport layer.
Winter 2008CS244a Handout 71 CS244a: An Introduction to Computer Networks Handout 7: Congestion Control Nick McKeown Professor of Electrical Engineering.
TTM1: Approaches to Optical Internet Packet Switching David K. Hunter and Ivan Andonovic.
1 Routing and Resilience in Future Optical Broadband Telecommunications Networks 21 st January 2004 Andrew S. T. Lee Supervisor: Dr. David Harle Broadband.
Chapter 11.4 END-TO-END ISSUES. Optical Internet Optical technology Protocol translates availability of gigabit bandwidth in user-perceived QoS.
Prentice HallHigh Performance TCP/IP Networking, Hassan-Jain TCP/IP Performance over Optical Networks.
Optimization Problems in Wireless Coding Networks Alex Sprintson Computer Engineering Group Department of Electrical and Computer Engineering.
A Comparison of RaDiO and CoDiO over IEEE WLANs May 25 th Jeonghun Noh Deepesh Jain A Comparison of RaDiO and CoDiO over IEEE WLANs.
© Janice Regan, CMPT 128, CMPT 371 Data Communications and Networking Congestion Control 0.
CS640: Introduction to Computer Networks Aditya Akella Lecture 15 TCP Congestion Control.
Tango1 Considering End-to-End QoS Constraints in IP Network Design and Planning M.Ajmone Marsan, M. Garetto, E. Leonardi. M. Mellia, E. Wille Dipartimento.
Network Processing Systems Design
Improving OBS Efficiency Li, Shuo, Meiqian Wang. Eric W. M. Wong, Moshe Zukerman City University of Hong Kong 1.
Chapter 3 outline 3.1 transport-layer services
Chapter 3 outline 3.1 Transport-layer services
SEMINAR ON Optical Burst Switching
Open Issues in Router Buffer Sizing
Lecture 19 – TCP Performance
OPTICAL PACKET SWITCHING
COMP/ELEC 429/556 Introduction to Computer Networks
Optical Burst Switching
TCP Overview.
Transport Layer: Congestion Control
Congestion Michael Freedman COS 461: Computer Networks
Presentation transcript:

TCP over OBS Joint Work with F. Baccelli, D. Hong, G. Petit, F. Poppe Julien Reynier Article available at

Capacity requirements of common services — 2 All rights reserved © 2002, Alcatel Overview > Introduction > Optical Burst Switching (OBS) : description > FDL > Optimization problem > The formula > Conclusion

Capacity requirements of common services — 3 All rights reserved © 2002, Alcatel Goal > Understand the interaction between TCP and physical layers for data transmission over OBS > OBS : Optical Burst Switching TCP layer IP layer OBS layer opticalelectrical

Capacity requirements of common services — 4 All rights reserved © 2002, Alcatel Overview > Introduction > Optical Burst Switching (OBS) : description > FDL > Optimization problem > The formula > Conclusion

Capacity requirements of common services — 5 All rights reserved © 2002, Alcatel Situation Optical Burst Switching : Optimize burst size Burstification of IP packets into Data Bursts (DBs) IP router OBS edge router OBS core router DWDM link Electronic processing of Burst Header Packets (BHPs) IP packets BHP Channel Group (CG) control (CCG) data (DCG)

Capacity requirements of common services — 6 All rights reserved © 2002, Alcatel Situation Optical Burst Switching IP router OBS Ingress edge router OBS core router DBs are routed in the optical domain IP packets DB Bursts disassembly

Capacity requirements of common services — 7 All rights reserved © 2002, Alcatel Specificity of OBS > At the ingress edge router, IP packets are assembled in lrager processing units, namely the Data Bursts (DBs). The loss of a single DB is the loss of many TCP-IP packets at the same time. > In the core routers, bursts wait in a queue implemented with optical fibers : the Fiber Delay Lines (FDLs). Bursts are not released as soon as the link becomes ready. – “Classical” queuing models do not work. Certain methods as Void Filling mitigate this effect. – The drawback is a complex architecture and a possible reordering of bursts, therefore of IP packets. Classical queue FDL length D

Capacity requirements of common services — 8 All rights reserved © 2002, Alcatel Overview > Introduction > Optical Burst Switching (OBS) : description > FDL > Optimization problem > The formula > Conclusion

Capacity requirements of common services — 9 All rights reserved © 2002, Alcatel FDL and a classical queue : > Without FDL : > With FDL : > The Void size is :

Capacity requirements of common services — 10 All rights reserved © 2002, Alcatel Voids TnTn T n+1 T n +kDT n+1 +k’D input : output : nn VnVn

Capacity requirements of common services — 11 All rights reserved © 2002, Alcatel Load factor for an FDL with infinite capacity

Capacity requirements of common services — 12 All rights reserved © 2002, Alcatel Example of an FDL workload Workload distribution of an FDL queue and for an M/M/1 queue with the same equivalent load D=0.1 hpkt  =7.8 1/hpkt =1 hpkt/s 1 hpkt=100 IP- packets  ’=95%

Capacity requirements of common services — 13 All rights reserved © 2002, Alcatel Loss rate due to buffer overflow with FDL b=1/  B=8D =1/b’ B:buffer D:delay line b:burst size b’:burst size + void size The loss rate at a “congestion epoch” is affine for a large range of values.

Capacity requirements of common services — 14 All rights reserved © 2002, Alcatel Overview > Introduction > Optical Burst Switching (OBS) : description > FDL > Optimization problem > The formula > Conclusion

Capacity requirements of common services — 15 All rights reserved © 2002, Alcatel Notation N sources TCP- controlled Link capacity C Queue size B RTT R Bottleneck router > N users share a bottleneck router with capacity C (Mb/s) with queue size B (byte), their RTT being R (ms).

Capacity requirements of common services — 16 All rights reserved © 2002, Alcatel Small bursts or large bursts ? > For large bursts Lower bandwidth waste due to guard bands – OBS needs time to reconfigure its switching array Lower processing time inside nodes > For small bursts Lower synchronization between sources due to TCP Lower burst formation time

Capacity requirements of common services — 17 All rights reserved © 2002, Alcatel Synchronization - AIMD > TCP is AIMD : Additive Increase, Multiplicative decrease > The proportion of sources that lose packets at that time is the synchronization X(t)X(t) t X ( t ) = throughput seen by a user T i = congestion epochs in the network TiTi T i-2 T2T2 TETE T1T1 TSTS Congestion avoidance: additive increase, multiplicative decrease T3T3 T i-1 T i-3 Slow startAdditiveIncreaseMultiplicativeDecrease

Capacity requirements of common services — 18 All rights reserved © 2002, Alcatel Overview > Introduction > Optical Burst Switching (OBS) : description > FDL > Optimization problem > The formula > Conclusion

Capacity requirements of common services — 19 All rights reserved © 2002, Alcatel AIMD framework The goodput is : The synchronization p(b,D) is : The AIMD framework models the impact of losses on the TCP- bandwidth. AIMD : “The AIMD model” : Infocom ‘02 by F. Baccelli et D. Hong

Capacity requirements of common services — 20 All rights reserved © 2002, Alcatel Formula  The study of FDL allows one to deduct the ratio of lost busts from the bandwidth.  This leads to a simple formula : For X=B/2 and Y=1/2

Capacity requirements of common services — 21 All rights reserved © 2002, Alcatel Overview > Introduction > Optical Burst Switching (OBS) : description > FDL > Optimization problem > The formula > Conclusion

Capacity requirements of common services — 22 All rights reserved © 2002, Alcatel Our results > Model FDL > Determine impact of burst size on aggregate TCP throughput the impact of FDL on synchronization of TCP sources the AIMD model allows one to find the bandwidth. > Determine the optimal burst size maximize the bandwidth. minimize the bandwidth waste due to OBS (Burst Header Packets, voids, guard bands).

Capacity requirements of common services — 23 All rights reserved © 2002, Alcatel Further research > A better representation of the increase of RTT due to burstification > Model Void Filling > Core routers with wavelength converters

Capacity requirements of common services — 25 All rights reserved © 2002, Alcatel Backup slides