Distributed Admission Control and Congestion Pricing Peter Key

Slides:



Advertisements
Similar presentations
Ch. 12 Routing in Switched Networks
Advertisements

Martin Suchara, Ryan Witt, Bartek Wydrowski California Institute of Technology Pasadena, U.S.A. TCP MaxNet Implementation and Experiments on the WAN in.
Peter Key Microsoft Research, Cambridge Fairness, Flow Control & Multi-User Games.
Congestion Control and Fairness Models Nick Feamster CS 4251 Computer Networking II Spring 2008.
Technische universiteit eindhoven 1 Problem 16: Design-space Exploration Jeroen Voeten, Bart Theelen Eindhoven University of Technology Embedded Systems.
1 EL736 Communications Networks II: Design and Algorithms Class1: Introduction Yong Liu 09/05/2007.
Game Theory and the Internet Flow and Congestion Control.
Min-Plus Linear Systems Theory and Bandwidth Estimation Min-Plus Linear Systems Theory and Bandwidth Estimation TexPoint fonts used in EMF. Read the TexPoint.
Ch. 12 Routing in Switched Networks Routing in Packet Switched Networks Routing Algorithm Requirements –Correctness –Simplicity –Robustness--the.
One More Bit Is Enough Yong Xia, RPI Lakshmi Subramanian, UCB Ion Stoica, UCB Shiv Kalyanaraman, RPI SIGCOMM’ 05, Philadelphia, PA 08 / 23 / 2005.
Traffic and routing. Network Queueing Model Packets are buffered in egress queues waiting for serialization on line Link capacity is C bps Average packet.
1 CONGESTION CONTROL. 2 Congestion Control When one part of the subnet (e.g. one or more routers in an area) becomes overloaded, congestion results. Because.
TELE202 Lecture 8 Congestion control 1 Lecturer Dr Z. Huang Overview ¥Last Lecture »X.25 »Source: chapter 10 ¥This Lecture »Congestion control »Source:
Congestion Control Algorithms
William Stallings Data and Computer Communications 7 th Edition Chapter 13 Congestion in Data Networks.
24-1 Chapter 24. Congestion Control and Quality of Service (part 1) 23.1 Data Traffic 23.2 Congestion 23.3 Congestion Control 23.4 Two Examples.
Architectures for Congestion-Sensitive Pricing of Network Services Thesis Defense by Murat Yuksel CS Department, RPI July 3 rd, 2002.
1.  Congestion Control Congestion Control  Factors that Cause Congestion Factors that Cause Congestion  Congestion Control vs Flow Control Congestion.
Engineering Internet QoS
Abhay.K.Parekh and Robert G.Gallager Laboratory for Information and Decision Systems Massachusetts Institute of Technology IEEE INFOCOM 1992.
Cloud Control with Distributed Rate Limiting Raghaven et all Presented by: Brian Card CS Fall Kinicki 1.
Resource pricing and the evolution of congestion control By R. J. Gibbens and F. P. Kelly.
One More Bit Is Enough Yong Xia, RPI Lakshminarayanan Subramanian, UCB Ion Stoica, UCB Shivkumar Kalyanaraman, RPI SIGCOMM’05, August 22-26, 2005, Philadelphia,
Congestion Control Tanenbaum 5.3, /12/2015Congestion Control (A Loss Based Technique: TCP)2 What? Why? Congestion occurs when –there is no reservation.
Explicit Congestion Notification ECN Tilo Hamann Technical University Hamburg-Harburg, Germany.
Congestion Pricing Overlaid on Edge-to-Edge Congestion Control Murat Yuksel, Shivkumar Kalyanaraman and Anuj Goel Rensselaer Polytechnic Institute, Troy,
Multiple constraints QoS Routing Given: - a (real time) connection request with specified QoS requirements (e.g., Bdw, Delay, Jitter, packet loss, path.
TCP Stability and Resource Allocation: Part I. References The Mathematics of Internet Congestion Control, Birkhauser, The web pages of –Kelly, Vinnicombe,
1 Random Early Detection Gateways for Congestion Avoidance Sally Floyd and Van Jacobson, IEEE Transactions on Networking, Vol.1, No. 4, (Aug 1993), pp
1 Internet Networking Spring 2003 Tutorial 11 Explicit Congestion Notification (RFC 3168)
1 End-to-End Detection of Shared Bottlenecks Sridhar Machiraju and Weidong Cui Sahara Winter Retreat 2003.
1 Auction or Tâtonnement – Finding Congestion Prices for Adaptive Applications Xin Wang Henning Schulzrinne Columbia University.
Distributed-Dynamic Capacity Contracting: A congestion pricing framework for Diff-Serv Murat Yuksel and Shivkumar Kalyanaraman Rensselaer Polytechnic Institute,
Random Early Detection Gateways for Congestion Avoidance
Combining Multipath Routing and Congestion Control for Robustness Peter Key.
1 A State Feedback Control Approach to Stabilizing Queues for ECN- Enabled TCP Connections Yuan Gao and Jennifer Hou IEEE INFOCOM 2003, San Francisco,
Advanced Computer Networks : RED 1 Random Early Detection Gateways for Congestion Avoidance Sally Floyd and Van Jacobson, IEEE Transactions on Networking,
Flow Models and Optimal Routing. How can we evaluate the performance of a routing algorithm –quantify how well they do –use arrival rates at nodes and.
A Generalized Processor Sharing Approach to Flow Control in Integrated Services Networks: The Single-Node Case Abhay K. Parekh, Member, IEEE, and Robert.
Courtesy: Nick McKeown, Stanford 1 TCP Congestion Control Tahir Azim.
1 Kommunikatsiooniteenuste arendus IRT0080 Loeng 7 Avo Ots telekommunikatsiooni õppetool, TTÜ raadio- ja sidetehnika inst.
“A non parametric estimate of performance in queueing models with long-range correlation, with applications to telecommunication” Pier Luigi Conti, Università.
جلسه دهم شبکه های کامپیوتری به نــــــــــــام خدا.
Univ. of TehranAdv. topics in Computer Network1 Advanced topics in Computer Networks University of Tehran Dept. of EE and Computer Engineering By: Dr.
Peter Key Service Differentiation: Congestion Pricing, Brokers and Bandwidth Futures.
Performance evaluation of video transcoding and caching solutions in mobile networks Jim Roberts (IRT-SystemX) joint work with Salah Eddine Elayoubi (Orange.
ACN: RED paper1 Random Early Detection Gateways for Congestion Avoidance Sally Floyd and Van Jacobson, IEEE Transactions on Networking, Vol.1, No. 4, (Aug.
FAIR CHARGES FOR INTERNET CONGESTION Damon Wischik Statistical Laboratory, Cambridge Electrical Engineering, Stanford
Congestion Control in CSMA-Based Networks with Inconsistent Channel State V. Gambiroza and E. Knightly Rice Networks Group
Peter Key Cambridge UK joint work with Richard Gibbens, Statistical Laboratory, Cambridge Uni. UK The.
TCP Trunking: Design, Implementation and Performance H.T. Kung and S. Y. Wang.
Queueing and Active Queue Management Aditya Akella 02/26/2007.
Multiplicative Wavelet Traffic Model and pathChirp: Efficient Available Bandwidth Estimation Vinay Ribeiro.
V. Fodor and Gy. Dan, KTH - Marholmen 2002 End-to-end control for audio-visual communication Viktoria Fodor and György Dán Laboratory for Communication.
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.
Jennifer Rexford Fall 2014 (TTh 3:00-4:20 in CS 105) COS 561: Advanced Computer Networks TCP.
Winter 2008CS244a Handout 71 CS244a: An Introduction to Computer Networks Handout 7: Congestion Control Nick McKeown Professor of Electrical Engineering.
CONGESTION CONTROL.
TCP continued. Discussion – TCP Throughput TCP will most likely generate the saw tooth type of traffic. – A rough estimate is that the congestion window.
Indian Institute of Technology Bombay 1 Communication Networks Prof. D. Manjunath
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.
ECEN 619, Internet Protocols and Modeling Prof. Xi Zhang Random Early Detection Gateways for Congestion Avoidance Sally Floyd and Van Jacobson, IEEE Transactions.
An End-to-End Service Architecture r Provide assured service, premium service, and best effort service (RFC 2638) Assured service: provide reliable service.
Providing QoS in IP Networks
Optimization-based Cross-Layer Design in Networked Control Systems Jia Bai, Emeka P. Eyisi Yuan Xue and Xenofon D. Koutsoukos.
Topics discussed in this section:
CONGESTION CONTROL.
The Network Layer Network Layer Design Issues:
Congestion Control, Quality of Service, & Internetworking
TCP Congestion Control
Presentation transcript:

Distributed Admission Control and Congestion Pricing Peter Key

Subplot … Can guarantees be provided using pricing alone? Refs: F.P. Kelly ( Stats Lab, Cambridge Uni. ), P.B Key, S.Zachary (Heriot-Watt Uni.), Distributed Admission Control, preprint

Outline zIntroduction yCongestion Pricing zAdaptive Admission Control yMathematical Framework yExamples xBinomial Model xVirtual Queue marking xCritical Timescales zDiscussion Commodity markets and Futures

Resource system (network) Resource j Capacity C j User /route r A jr links users to resources

Basic Idea zUsers generate load (packets) zNetwork sends back signals (load dependent) zSignals : proportional to load yAct as feedback indicators yRepresent pricing signals xmarginal incremental costs (derivatives …) xcongestion costs xreal money or virtual / distributed mint

Optimisation Framework (for fairness) System optimum U is utility User optimum C is cost function,eg

Solution Consistent set of taxes (prices) and load exist s.t. Eg Network chooses taxes, user chooses load, solution is network, user and System optimal. But dependent on Utility function, so ….

Matching prices to load zFor bounded prices, have to match price load to capacity zie, require maximum amount users prepared to pay < maximum network can charge zEg, if x r satisfies then require

Admission Control zSend a number of probe packets through the network zEnter the network if none of these packets are marked zAssume: Poisson arrivals, rate zLet a(m j ) be probability accepted at node j independently

User policy User /route r M probe packets Enter if less than m probe Packets marked

Product form distribution Equilibrium distribution for the number of calls in progress n-1 n+1 n v a(n-1) v a(n) n+1 n

Fixed Point Approximation zDefine stationary acceptance probability for J={j}, R={r} zThen fixed point approximation for network has unique solution

Acceptance Probabilities Example 1 zEg, let 1-a j (m j ) be probability any of a number of probe packets are marked zEg for a burst-scale model, where there is long- range dependence

Rejection Probabilities & PDFs Rejection probabilities Equilibrium CDF Setup: =50, thresholds 10, 20 n

PDFs Setup: =50, thresholds 10, 20 n Setup: =100, thresholds 10, 20

Shadow Prices (buffered) Max Queue Length Q Time Queue Length packets Q Fixed Service Rate Mark all packets from start of busy period until last packet loss.

Virtual Queue Marking zPut arrivals into a virtual queue, and mark on this zCapacity capacity c v c, eg c zBuffer size b v b, eg b

Virtual Queue Example zSuppose we want to track derivative of queue, (or suppose cost=P[exceed thresh) zM/M/1 (can use other SRD processes) zEquate derivate to a VQ with reduced rate zFor virtual queue, rate, thresh K-1, put

VQ Thresholds

Timescales Connection Reaction (RTT) Packet Level average rate Seconds line rate ms s ApplicationNetwork msms s Critical timescale

Critical Timescales zLarge deviation approach (many sources asymptotic – Courcoubetis and Weber) where t* and s* are extremals, t* is the critical timescale. If mark as shadow price, is typical marking time

Critical timescales for VQ zEg for a Gaussian process, arrival rate, Hurst parameter H

Critical timescales zExample 1

Critical timescales zExample 2 yBUT, need to have critical timescale less than time between arrivals (for decisions to be independent) yThis is (mean holding time)/(number of calls) in equilibrium 0 as n zHence, ykeep virtual queue small, just for cell scale

Congestion Prices (Timescales) 60 secs LAN) 1 Sec (Backbone)

Example 2 – packet marking zMark packets if size of Virtual Queue exceeds threshold zIfM probe packets sent zwhere is mean packet service time (if connections generate packets at rate r, service rate is c, then =r/c, and 1/ represents capacity of queue )

Rejection Probabilities & PDFs, VQ marking Rejection probabilities Equilibrium CDF Setup: =50, thresholds 5,10 n

PDFs Setup: =50, thresholds 5,10 n Setup: =100, thresholds 5,10

Blocking vs. Marking (price) VQ marking, threshold K=10, capacity (1/ )=100 1 Probe packet 5 Probe packets Blocking marking

Mixing adaptive and non- adaptive traffic zSimple model: two types of traffic yNon-adaptive traffic, requires unit bandwidth yAdaptive traffic: reacts to signals can halve its bandwidth requirement zSuppose price (congestion marking probability) not to go above 0.2 zGives acceptance boundaries

Price regions

Acceptance Boundaries

To give a price blocking of Capacity =25.5 (eg LAN with voice, PCM coding) Prop. Of adaptive traffic required Total arrival rate

Discussion zCongestion pricing works well for adaptive applications zWe have constructed a model for streams/flows where decisions made by end-systems zSystem is robust, and can be analysed /engineered

Facilitators zCritical timescales (of marking) small compared to interarrival times, (comparable to RTTs?) zSmall buffers in Virtual Queue (compared to transmission delay) to detect quickly zTarget loads below 100% … zSimple feedback signal, eg ECN bit/byte zSignal reflects costs zPrices need to match demand zUser interface simple (risk apportionment)