UCB Improvements in Core-Stateless Fair Queueing (CSFQ) Ling Huang U.C. Berkeley cml.me.berkeley.edu/~hlion.

Slides:



Advertisements
Similar presentations
Balaji Prabhakar Active queue management and bandwidth partitioning algorithms Balaji Prabhakar Departments of EE and CS Stanford University
Advertisements

Congestion Control and Fairness Models Nick Feamster CS 4251 Computer Networking II Spring 2008.
Using Edge-To-Edge Feedback Control to Make Assured Service More Assured in DiffServ Networks K.R.R.Kumar, A.L.Ananda, Lillykutty Jacob Centre for Internet.
Fair Queueing. Design space Buffer management: –RED, Drop-Tail, etc. Scheduling: which flow to service at a given time –FIFO –Fair Queueing.
RED-PD: RED with Preferential Dropping Ratul Mahajan Sally Floyd David Wetherall.
Computer Networking Lecture 20 – Queue Management and QoS.
FAST TCP Anwis Das Ajay Gulati Slides adapted from : IETF presentation slides Link:
WHITE – Achieving Fair Bandwidth Allocation with Priority Dropping Based on Round Trip Time Name : Choong-Soo Lee Advisors : Mark Claypool, Robert Kinicki.
CSIT560 Internet Infrastructure: Switches and Routers Active Queue Management Presented By: Gary Po, Henry Hui and Kenny Chong.
CS640: Introduction to Computer Networks Mozafar Bag-Mohammadi Lecture 3 TCP Congestion Control.
CS 268: Lecture 8 Router Support for Congestion Control Ion Stoica Computer Science Division Department of Electrical Engineering and Computer Sciences.
CS 4700 / CS 5700 Network Fundamentals Lecture 12: Router-Aided Congestion Control (Drop it like it’s hot) Revised 3/18/13.
Router-assisted congestion control Lecture 8 CS 653, Fall 2010.
Advanced Computer Networking Congestion Control for High Bandwidth-Delay Product Environments (XCP Algorithm) 1.
Ion Stoica, Scott Shenker, and Hui Zhang SIGCOMM’98, Vancouver, August 1998 subsequently IEEE/ACM Transactions on Networking 11(1), 2003, pp Presented.
Max Min Fairness How define fairness? “ Any session is entitled to as much network use as is any other ” ….unless some sessions can use more without hurting.
XCP: Congestion Control for High Bandwidth-Delay Product Network Dina Katabi, Mark Handley and Charlie Rohrs Presented by Ao-Jan Su.
Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 1 ECSE-6600: Internet Protocols Informal Quiz #05: SOLUTIONS Shivkumar Kalyanaraman: GOOGLE: “Shiv.
1 Core Stateless Fair Queueing Ion Stoica Hui Zhang Scott Shenker CMU CMU Xerox PARC CMU CMU Xerox PARC.
1 Core-Stateless Fair Queueing: A Scalable Architecture to Approximate Fair Bandwidth Allocations in High Speed Networks Core-Stateless Fair Queueing:
Nick McKeown CS244 Lecture 6 Packet Switches. What you said The very premise of the paper was a bit of an eye- opener for me, for previously I had never.
1 Minseok Kwon and Sonia Fahmy Department of Computer Sciences Purdue University {kwonm, All our slides and papers.
AQM for Congestion Control1 A Study of Active Queue Management for Congestion Control Victor Firoiu Marty Borden.
Diffusion Mechanisms for Active Queue Management Department of Electrical and Computer Engineering University of Delaware May 19th / 2004 Rafael Nunez.
High speed TCP’s. Why high-speed TCP? Suppose that the bottleneck bandwidth is 10Gbps and RTT = 200ms. Bandwidth delay product is packets (1500.
EE689 Lecture 5 Review of last lecture More on HPF RED.
CS 268: Lecture 8 (Router Support for Congestion Control) Ion Stoica February 19, 2002.
1 Core-Stateless Fair Queueing: Achieving Approximately Fair Bandwidth Allocations in High Speed Networks Ion Stoica,Scott Shenker, and Hui Zhang SIGCOMM’99,
1 Emulating AQM from End Hosts Presenters: Syed Zaidi Ivor Rodrigues.
Computer Networking Lecture 17 – Queue Management As usual: Thanks to Srini Seshan and Dave Anderson.
UCB Review for M2 Jean Walrand U.C. Berkeley
1 Core-Stateless Fair Queueing: Achieving Approximately Fair Bandwidth Allocations in High Speed Networks Ion Stoica,Scott Shenker, and Hui Zhang SIGCOMM’99,
Congestion Control for High Bandwidth-delay Product Networks Dina Katabi, Mark Handley, Charlie Rohrs.
Core Stateless Fair Queueing Stoica, Shanker and Zhang - SIGCOMM 98 Rigorous fair Queueing requires per flow state: too costly in high speed core routers.
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.
Core Stateless Fair Queueing Stoica, Shanker and Zhang - SIGCOMM 98 Fair Queueing requires per flow state: too costly in high speed core routers Yet, some.
1 Queue Management Hamed Khanmirza Principles of Networking University of Tehran.
Advance Computer Networking L-5 TCP & Routers Acknowledgments: Lecture slides are from the graduate level Computer Networks course thought by Srinivasan.
Understanding the Performance of TCP Pacing Amit Aggarwal, Stefan Savage, Thomas Anderson Department of Computer Science and Engineering University of.
ACN: CSFQ1 CSFQ Core-Stateless Fair Queueing Presented by Nagaraj Shirali Choong-Soo Lee ACN: CSFQ1.
Advance Computer Networking L-6 TCP & Routers Acknowledgments: Lecture slides are from the graduate level Computer Networks course thought by Srinivasan.
Advanced Computer Networking
ACN: RED paper1 Random Early Detection Gateways for Congestion Avoidance Sally Floyd and Van Jacobson, IEEE Transactions on Networking, Vol.1, No. 4, (Aug.
1 On Class-based Isolation of UDP, Short-lived and Long-lived TCP Flows by Selma Yilmaz Ibrahim Matta Computer Science Department Boston University.
27th, Nov 2001 GLOBECOM /16 Analysis of Dynamic Behaviors of Many TCP Connections Sharing Tail-Drop / RED Routers Go Hasegawa Osaka University, Japan.
High-speed TCP  FAST TCP: motivation, architecture, algorithms, performance (by Cheng Jin, David X. Wei and Steven H. Low)  Modifying TCP's Congestion.
Presented by: Peng Wang EE Department University of Delaware A Probabilistic Approach for Achieving Fair Bandwidth Allocation in CSFQ.
Queueing and Active Queue Management Aditya Akella 02/26/2007.
9.7 Other Congestion Related Issues Outline Queuing Discipline Avoiding Congestion.
15744 Course Project1 Evaluation of Queue Management Algorithms Ningning Hu, Liu Ren, Jichuan Chang 30 April 2001.
1 Analysis of a window-based flow control mechanism based on TCP Vegas in heterogeneous network environment Hiroyuki Ohsaki Cybermedia Center, Osaka University,
1 Core-Stateless Fair Queueing: A Scalable Architecture to Approximate Fair Bandwidth Allocations in High Speed Networks Core-Stateless Fair Queueing:
Explicit Allocation of Best-Effort Service Goal: Allocate different rates to different users during congestion Can charge different prices to different.
XCP: eXplicit Control Protocol Dina Katabi MIT Lab for Computer Science
Spring Computer Networks1 Congestion Control Sections 6.1 – 6.4 Outline Preliminaries Queuing Discipline Reacting to Congestion Avoiding Congestion.
1 Sheer volume and dynamic nature of video stresses network resources PIE: A lightweight latency control to address the buffer problem issue Rong Pan,
Univ. of TehranIntroduction to Computer Network1 An Introduction Computer Networks An Introduction to Computer Networks University of Tehran Dept. of EE.
1 Lecture 15 Internet resource allocation and QoS Resource Reservation Protocol Integrated Services Differentiated Services.
Corelite Architecture: Achieving Rated Weight Fairness
Topics discussed in this section:
CS 268: Lecture 6 Scott Shenker and Ion Stoica
Core-Stateless Fair Queueing: A Scalable Architecture to Approximate Fair Bandwidth Allocations in High Speed Networks Ion Stoica, Scott Shenker, and Hui.
Columbia University in the city of New York
EE 122: Router Support for Congestion Control: RED and Fair Queueing
Advance Computer Networking
Advance Computer Networking
Presentation transcript:

UCB Improvements in Core-Stateless Fair Queueing (CSFQ) Ling Huang U.C. Berkeley cml.me.berkeley.edu/~hlion

UCB Achievements Analyze the limitation of current approach Failure to work with congestion avoidance algorithm. Heavy dependence on the estimation algorithms Estimation has big deviation when many flows startup simultaneously and during bursty traffic. Achieve three improvements Achieving fair share when working with congestion avoidance algorithm. Application of high order Low-Pass-Filter (LPF) in flow arrival rate estimation. Application of control-theory in fair share estimation.

UCB Algorithm of CSFQ Edge routers put flow state in packet‘s header. Core routers estimate fair share and drop. incoming packets with probability of A flow should get bandwidth. Core router updates fair share  as follows: if (A > C)  new =  old * C / F else  new = max (r i ), where r i  active flows Combining fair share computation and probabilistic dropping to approximate fair queueing!

UCB Improv.1: CSFQ working with TCP Vegas CSFQ get fare share to incoming flows, but Not accurately approximate delay properties of Fair Queueing. Incompatible with congestion avoidance mechanisms. Fig 1. CSFQ work with TCP Reno Fig 2. CSFQ work with TCP Vegas

UCB Improv. 1: CSFQ working with TCP Vegas High priority queue in core router Flows whose rates are less than their fare share go into high priority queue and get service first. Restores fairness when work with TCP Vegas. Fig 3. CSFQ + Priority Queue Fig 4. CSFQ + Priority Queue work with TCP Reno work with TCP Vegas

UCB Improv. 2: LPF in arrival rate estimation Current Approach employ 1 st order Low-Pass-Filter: Better properties from 2 nd order Low-Pass-Filter: Fast response to burst traffic. Estimation result more smooth.

UCB Improv. 2: LPF in arrival rate estimation Fast response to burst traffic Input singal Output of 1st order LPF Output of 2nd order LPF Followings are the same Fig 5. Response to pulse signal

UCB Improv. 2: LPF in arrival rate estimation Rate estimation for arrival traffic Fig 6. Estimation of arrival rate for UDP(1Mbps) Fig 7. Estimation of arrival rate for TCP (0.5Mbps)

UCB Improv. 2: LPF in arrival rate estimation Results of 2 nd Low-Pass-Filter in CSFQ Service allocated to flows more fare. Fig 8. CSFQ + Priority Queue + 2 nd LPF Fig 9. CSFQ + Priority Queue + 2 nd LPF work with TCP Reno work with TCP Vegas

UCB Improv. 3: Applying control theory in CSFQ Picture: Traffic model faucet and Flow 1 Flow n Exponential Average Exponential Average Arrival Rate Arrival Rate Estimating accepting rate Packet Dropper Observer R C s Q 0 Q Fig 10. Architecture of CSFQ scheduling: the same model as that in a water reservation system R C s Q 0 Q

UCB Improv. 3: Applying control theory in CSFQ Dynamic equation: using queue occupancy information to compute arrival rate. R: the aggregate accepted traffic during one update interval. C s : the output link capacity. Q: the buffer occupancy at current time. Q 0 : equilibrium point that we want the buffer occupancy to be. R C s Q 0 Q

UCB Improv. 3: Applying control theory in CSFQ Use queue occupancy information to predict the next allowed accepting rate: flow and congestion control. Allocate the allowed accepting rate to the incoming flows, fair share is the unique solution of equation:

UCB Results Improv. 3: Applying control theory in CSFQ Fig 11. CSFQ + control theory Fig 12. CSFQ + control theory work with TCP work with TCP Reno

UCB Results (cont.) Improv. 3: Applying control theory in CSFQ Fig 13. CSFQ + control theory work with TCP Vega Possible reason: priority queue confusing control system with buffer length. Fig 14. Average throughput by a TCP sharing a link of capacity 10 Mbps with (n-1) UDP flows. ( Import from [Hoon-Tong’00] )

UCB Conclusion: control theory produce stable and robust system. Without the the estimation of aggregate arrival rate and accepted traffic. Compute fare share continuously. Exhibit good transient and steady behavior. Achieve high utilization. Improv. 3: Applying control theory in CSFQ

UCB Three improvements in CSFQ High priority helps CSFQ achieve fair share when working with congestion avoidance algorithm. High order Low-Pass-Filter (LPF) helps improve fair share rate. Control-theory produce robust and highly utilized system. CS268 is a productive class Paper review and lecture let me know every aspect of network. Class project help me go deep into algorithm of resource allocation and packet scheduling. Work hard, you get achievement! Summary

UCB CSFQ work with TCP Renog CSFQ + Priority Queue work with TCP Reno CSFQ + Priority Queue + 2 nd LPF CSFQ + control theory work with TCP Reno work with TCP Reno

UCB CSFQ work with TCP Vegas CSFQ + Priority Queue work with TCP Vegas CSFQ + Priority Queue + 2 nd LPF CSFQ + control theory work with TCP Vegas work with TCP Vegas