1 Time-scale Decomposition and Equivalent Rate Based Marking Yung Yi, Sanjay Shakkottai ECE Dept., UT Austin Supratim Deb.

Slides:



Advertisements
Similar presentations
Helping TCP Work at Gbps Cheng Jin the FAST project at Caltech
Advertisements

Internet Protocols Steven Low CS/EE netlab.CALTECH.edu October 2004 with J. Doyle, L. Li, A. Tang, J. Wang.
Restless bandits and congestion control Mark Handley, Costin Raiciu, Damon Wischik UCL.
CSIT560 Internet Infrastructure: Switches and Routers Active Queue Management Presented By: Gary Po, Henry Hui and Kenny Chong.
On Large-Scale Peer-to-Peer Streaming Systems with Network Coding Chen Feng, Baochun Li Dept. of Electrical and Computer Engineering University of Toronto.
Simulating Large Networks using Fluid Flow Model Yong Liu Joint work with Francesco LoPresti, Vishal Misra Don Towsley, Yu Gu.
REM : Active Queue Management Sanjeewa Athuraliya, Victor H. Li Steven H. Low, Qinghe Yin Presented by Hwangnam Kim.
Advanced Computer Networking Congestion Control for High Bandwidth-Delay Product Environments (XCP Algorithm) 1.
Congestion Control An Overview -Jyothi Guntaka. Congestion  What is congestion ?  The aggregate demand for network resources exceeds the available capacity.
XCP: Congestion Control for High Bandwidth-Delay Product Network Dina Katabi, Mark Handley and Charlie Rohrs Presented by Ao-Jan Su.
TCP Stability and Resource Allocation: Part II. Issues with TCP Round-trip bias Instability under large bandwidth-delay product Transient performance.
Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 1 ECSE-6600: Internet Protocols Informal Quiz #05: SOLUTIONS Shivkumar Kalyanaraman: GOOGLE: “Shiv.
Control Theory in TCP Congestion Control and new “FAST” designs. Fernando Paganini and Zhikui Wang UCLA Electrical Engineering July Collaborators:
1 Minseok Kwon and Sonia Fahmy Department of Computer Sciences Purdue University {kwonm, All our slides and papers.
Lecture 9. Unconstrained Optimization Need to maximize a function f(x), where x is a scalar or a vector x = (x 1, x 2 ) f(x) = -x x 2 2 f(x) = -(x-a)
Comparing flow-oblivious and flow-aware adaptive routing Sara Oueslati and Jim Roberts France Telecom R&D CISS 2006 Princeton March 2006.
Charge-Sensitive TCP and Rate Control Richard J. La Department of EECS UC Berkeley November 22, 1999.
Bandwidth sharing: objectives and algorithms Jim Roberts France Télécom - CNET Laurent Massoulié Microsoft Research.
TCP Stability and Resource Allocation: Part I. References The Mathematics of Internet Congestion Control, Birkhauser, The web pages of –Kelly, Vinnicombe,
Diffusion Mechanisms for Active Queue Management Department of Electrical and Computer Engineering University of Delaware May 19th / 2004 Rafael Nunez.
Network Bandwidth Allocation (and Stability) In Three Acts.
High speed TCP’s. Why high-speed TCP? Suppose that the bottleneck bandwidth is 10Gbps and RTT = 200ms. Bandwidth delay product is packets (1500.
High-performance bulk data transfers with TCP Matei Ripeanu University of Chicago.
1 Minseok Kwon and Sonia Fahmy Department of Computer Sciences Purdue University {kwonm, TCP Increase/Decrease.
Heterogeneous Congestion Control Protocols Steven Low CS, EE netlab.CALTECH.edu with A. Tang, J. Wang, D. Wei, Caltech M. Chiang, Princeton.
A Real-Time Video Multicast Architecture for Assured Forwarding Services Ashraf Matrawy, Ioannis Lambadaris IEEE TRANSACTIONS ON MULTIMEDIA, AUGUST 2005.
1 Emulating AQM from End Hosts Presenters: Syed Zaidi Ivor Rodrigues.
ACN: Congestion Control1 Congestion Control and Resource Allocation.
Congestion Control for High Bandwidth-delay Product Networks Dina Katabi, Mark Handley, Charlie Rohrs.
Rafael C. Nunez - Gonzalo R. Arce Department of Electrical and Computer Engineering University of Delaware May 19 th, 2005 Diffusion Marking Mechanisms.
CS :: Fall 2003 TCP Friendly Streaming Ketan Mayer-Patel.
Diffusion Mechanisms for Active Queue Management Department of Electrical and Computer Engineering University of Delaware May 19th / 2004 Rafael Nunez.
Congestion Control for High Bandwidth-Delay Product Environments Dina Katabi Mark Handley Charlie Rohrs.
1 A State Feedback Control Approach to Stabilizing Queues for ECN- Enabled TCP Connections Yuan Gao and Jennifer Hou IEEE INFOCOM 2003, San Francisco,
Diffusion Early Marking Department of Electrical and Computer Engineering University of Delaware May / 2004 Rafael Nunez Gonzalo Arce.
Buffer requirements for TCP: queueing theory & synchronization analysis Gaurav RainaDamon Wischik CambridgeUCL.
Congestion models for bursty TCP traffic Damon Wischik + Mark Handley University College London DARPA grant W911NF
TFRC: TCP Friendly Rate Control using TCP Equation Based Congestion Model CS 218 W 2003 Oct 29, 2003.
ACN: RED paper1 Random Early Detection Gateways for Congestion Avoidance Sally Floyd and Van Jacobson, IEEE Transactions on Networking, Vol.1, No. 4, (Aug.
Korea Advanced Institute of Science and Technology Network Systems Lab. 1 Dual-resource TCP/AQM for processing-constrained networks INFOCOM 2006, Barcelona,
27th, Nov 2001 GLOBECOM /16 Analysis of Dynamic Behaviors of Many TCP Connections Sharing Tail-Drop / RED Routers Go Hasegawa Osaka University, Japan.
Measurement, Modeling and Analysis of the Internet Wang Xiaofei Vishal Misra, Columbia University.
High-speed TCP  FAST TCP: motivation, architecture, algorithms, performance (by Cheng Jin, David X. Wei and Steven H. Low)  Modifying TCP's Congestion.
Acknowledgments S. Athuraliya, D. Lapsley, V. Li, Q. Yin (UMelb) S. Adlakha (UCLA), J. Doyle (Caltech), K. Kim (SNU/Caltech), F. Paganini (UCLA), J. Wang.
TCP with Variance Control for Multihop IEEE Wireless Networks Jiwei Chen, Mario Gerla, Yeng-zhong Lee.
Murari Sridharan and Kun Tan (Collaborators: Jingmin Song, MSRA & Qian Zhang, HKUST.
Hybrid Modeling of TCP Congestion Control João P. Hespanha, Stephan Bohacek, Katia Obraczka, Junsoo Lee University of Southern California.
Active Queue Management in Internet and Wireless Networks X. Deng, S. Yi, G. Kesidis and C. R. Das The Pennsylvania State University Stabilized queue size.
Transport Layer 3-1 Chapter 3 Transport Layer Computer Networking: A Top Down Approach 6 th edition Jim Kurose, Keith Ross Addison-Wesley March
Winter 2008CS244a Handout 81 CS244a: An Introduction to Computer Networks Handout 8: Congestion Avoidance and Active Queue Management Nick McKeown Professor.
We used ns-2 network simulator [5] to evaluate RED-DT and compare its performance to RED [1], FRED [2], LQD [3], and CHOKe [4]. All simulation scenarios.
Scalable Laws for Stable Network Congestion Control Fernando Paganini UCLA Electrical Engineering IPAM Workshop, March Collaborators: Steven Low,
XCP: eXplicit Control Protocol Dina Katabi MIT Lab for Computer Science
Analysis and Design of an Adaptive Virtual Queue (AVQ) Algorithm for AQM By Srisankar Kunniyur & R. Srikant Presented by Hareesh Pattipati.
Queueing theory for TCP Damon Wischik, UCL Gaurav Raina, Cambridge.
1 Transport Bandwidth Allocation 3/29/2012. Admin. r Exam 1 m Max: 65 m Avg: 52 r Any questions on programming assignment 2 2.
1 Network Transport Layer: TCP Analysis and BW Allocation Framework Y. Richard Yang 3/30/2016.
Internet Networking recitation #9
Chapter 3 outline 3.1 transport-layer services
CUBIC Marcos Vieira.
Router-Assisted Congestion Control
Congestion Control and Resource Allocation
Columbia University in the city of New York
Lecture 19 – TCP Performance
Random Early Detection Gateways for Congestion Avoidance
RAP: Rate Adaptation Protocol
Stability of Congestion Control Algorithms Using Control Theory with an application to XCP Ioannis Papadimitriou George Mavromatis.
Understanding Congestion Control Mohammad Alizadeh Fall 2018
Congestion Control and Resource Allocation
Queueing Problem The performance of network systems rely on different delays. Propagation/processing/transmission/queueing delays Which delay is affected.
Presentation transcript:

1 Time-scale Decomposition and Equivalent Rate Based Marking Yung Yi, Sanjay Shakkottai ECE Dept., UT Austin Supratim Deb LIDS, MIT

2 Contents Introduction Marking Based Congestion Control, Motivation System Model and Problem Definition Source Update Model: Congestion Control Algorithm Intuition and Results Simulation Results Summary

3 Marking Based Congestion Control Router reacts to the aggregate flow passing through it Marks packets during congestion control Explicit Congestion Notification (ECN) [Floyd 94] Active Queue Management (AQM) [Kelly 98, Kunniyur 01, Low 99, Towsley 00] Users adapt their transmission rate Congestion Control System Marking function at routers Rate Adaptation algorithm at sources marked packetunmarked packet Source decreases rate Source increases rate

4 How Do Routers Mark Packets? 1 0 total arrival rate queue length Rate Based Marking Queue Based Marking Marking probability Adjust its transmission rate depending on volume of marks received marked packetunmarked packet Source decreases rate Source increases rate

5 How Can We Simulate the Internet? Pure Packet Model: Discrete Event Simulation [ns2, pdns, parsec, ssfnet] Accurate transient behavior, but high complexity State changes at discrete events (message generated, packet arrival, packet departure, etc.) Computation: a sequence of event computations, processed in time stamp order Most of complexity: Queueing Dynamics Pure Fluid Model [Danzig 96, Towsley 00, 04, Hou 04, others] Fast and low complexity, but only steady state and approximate results Time-stepped evolution of system states Good in parallel processing slow, accurate, off-line fast, approximate, on-line

6 Motivation In reality, A significant number of uncontrolled flows (e.g. multimedia and web mice) Queue based marking (e.g., REM and RED) is popular in the real implementation cf) REM: Random Exponential Marking, RED: Random Early Detection Question 1: Can queue dynamics be decoupled from user dynamics? Question 2: What is the implication on the marking function? Is there an equivalent marking function which depends only on “instantaneous” data transmission rate?

7 Contents Introduction Marking Based Congestion Control, Motivation System Model and Problem Definition Source Update Model: Congestion Control Algorithm Intuition and Results Simulation Results Summary

8 System Model n controlled flows, n uncontrolled flows Controlled flows Differential equation based controller with queue based marking Link Bandwidth: n c Capacity proportional to the number of flows Small Buffer Regime

9 Modeling of Buffer Size: nB or B ? Queue buffer scale linearly with the # of flows or not ? Small Buffer Regime High Link speeds  need high-speed buffer with high cost Buffers need not scale with the link speed in order to achieve significant multiplexing gain [Cao & Ramanan 02] [Mandjes & Kim 01] [Mckeown 04]

10 What Kind of Source Controller Model? [kelly 98, et el] Rate based Marking Queue Based Marking Uncontrolled rate Controlled rate : Utility function of i-th controlled flow : TCP controller : Proportional Fair Controller Problem in this research: Finding given

11 Optimization Framework [Kelly et el.] c1c1 c2c2 x1x1 x2x2 x3x3 Differential Equation Based Distributed Congestion Control Algorithm Resource Constraints in Wired Networks

12 Contents Introduction Marking Based Congestion Control, Motivation System Model and Problem Definition Source Update Model: Congestion Control Algorithm Intuition and Results Simulation Results Summary

13 Intuition large number of uncontrolled flows (e.g., multimedia or web mice) large amount of randomness ……. Q length 1 round trip time Controlled flows (e.g., TCP flows) End-system controller influenced only through the (statistical) stationary queueing dynamics Queue Feedback (Ack) large number of cycles, where queue becomes empty Underlying Theory: Law of large numbers and Ergodic theorem

14 Large System Limit Unscaled system: n flows Uncontrolled flows: Stationary point process aggregate arrival rate: not necessarily a Poisson process Limiting system M/D/1 queue with service rate: n uncontrolled flows (aggregate arrival rate = ) n  suitable scaling Poisson( ) n controlled flows (aggregate arrival rate = )

15 Implications Low complexity model for large system dynamics No queueing dynamics in the model Simpler analysis and simulation Asynchronous event simulation  Synchronous time-stepped evolving simulation n  suitable scaling Queue Based Marking Function Rate Based Marking Function M/D/1 Queue Cf) Discrete Time Domain S. Deb and R. Srikant. Rate-based versus Queue-based models of congestion control. ACM Sigmetrics, June 2004.

16 Equivalent Rate Based Marking Equivalent Rate Based Marking Function x: arrival rate of controlled flows Lambda: arrival rate of uncontrolled flows Depends only on the stationary distribution of an M/D/1 queue

17 Sketch of Proof

18 Example : REM [Low 99] REM’s queue based marking function Equivalent Marking Function (from P-K formula)

19 Contents Introduction Marking Based Congestion Control, Motivation System Model and Problem Definition Source Update Model: Congestion Control Algorithm Intuition and Results Simulation Results Summary

20 Simulation Results (1) Bottleneck bw: 100 x n pkts n = 100 ( n: # of controlled and uncontrolled flows ) TCP Sack, Proportional Fair Controller REM, RED Queue based marking scheme

21 Simulation Results (2) Throughput Distribution of CWND

22 Summary In the Internet Significant number of uncontrolled (short and unresponsive) flows Queue based marking is popular Randomness due to short and unresponsive flows in the Internet sufficient to decouple the dynamics of the router queues from those of end controllers  We can find an equivalent rate based marking function given the queue based marking function Easier analysis and simulation We can apply nice mathematical tools to the analysis Asynchronous event-driven simulation  Synchronous fluid model based time-stepped evolving simulation  leading to low simulation complexity

23 References Y. Yi, S. Deb, and S. Shakkottai, “Short Queue Behavior and Rate Based Marking,” Proceedings of the 38th CISS, Princeton University, NJ, March, A longer version has been submitted to IEEE/ACM Transactions on Networking Cao and Ramanan, “A Poisson Limit for Buffer Overflow Probabilities,” Proc. IEEE Infocom, June, Daley and Vere-Jones, “An Introduction to the Theory of Point Processes,” Springer-Verlag, R. Srikant. "The Mathematics of Internet Congestion Control." Birkhauser, 2004.