Www.ict.csiro.au End-2-End QoS Internet Presented by: Zvi Rosberg 3 Dec, 2007 Caltech Seminar.

Slides:



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

Internet Protocols Steven Low CS/EE netlab.CALTECH.edu October 2004 with J. Doyle, L. Li, A. Tang, J. Wang.
Hot Sticky Random Multipath or Energy Pooling Jon Crowcroft,
Receiver-driven Layered Multicast S. McCanne, V. Jacobsen and M. Vetterli University of Calif, Berkeley and Lawrence Berkeley National Laboratory SIGCOMM.
Improving TCP Performance over Mobile Ad Hoc Networks by Exploiting Cross- Layer Information Awareness Xin Yu Department Of Computer Science New York University,
Congestion Control Created by M Bateman, A Ruddle & C Allison As part of the TCP View project.
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.
Receiver-driven Layered Multicast S. McCanne, V. Jacobsen and M. Vetterli SIGCOMM 1996.
CPSC Topics in Multimedia Networking A Mechanism for Equitable Bandwidth Allocation under QoS and Budget Constraints D. Sivakumar IBM Almaden Research.
Control Theory in TCP Congestion Control and new “FAST” designs. Fernando Paganini and Zhikui Wang UCLA Electrical Engineering July Collaborators:
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)
Self-Tuning End-2-End QoS Internet ICT Centre Presented by: Zvi Rosberg 21 March 2007.
Charge-Sensitive TCP and Rate Control Richard J. La Department of EECS UC Berkeley November 22, 1999.
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,
ACN: IntServ and DiffServ1 Integrated Service (IntServ) versus Differentiated Service (Diffserv) Information taken from Kurose and Ross textbook “ Computer.
TCP Congestion Control TCP sources change the sending rate by modifying the window size: Window = min {Advertised window, Congestion Window} In other words,
Heterogeneous Congestion Control Protocols Steven Low CS, EE netlab.CALTECH.edu with A. Tang, J. Wang, D. Wei, Caltech M. Chiang, Princeton.
EE 685 presentation Optimization Flow Control, I: Basic Algorithm and Convergence By Steven Low and David Lapsley Asynchronous Distributed Algorithm Proof.
Rethinking Internet Traffic Management: From Multiple Decompositions to a Practical Protocol Jiayue He Princeton University Joint work with Martin Suchara,
Presented by Anshul Kantawala 1 Anshul Kantawala FAST TCP: From Theory to Experiments C. Jin, D. Wei, S. H. Low, G. Buhrmaster, J. Bunn, D. H. Choe, R.
CSc 461/561 CSc 461/561 Multimedia Systems Part C: 3. QoS.
Congestion Control for High Bandwidth-delay Product Networks Dina Katabi, Mark Handley, Charlie Rohrs.
Multipath Protocol for Delay-Sensitive Traffic Jennifer Rexford Princeton University Joint work with Umar Javed, Martin Suchara, and Jiayue He
Congestion Control for High Bandwidth-Delay Product Environments Dina Katabi Mark Handley Charlie Rohrs.
UCB Improvements in Core-Stateless Fair Queueing (CSFQ) Ling Huang U.C. Berkeley cml.me.berkeley.edu/~hlion.
Not All Microseconds are Equal: Fine-Grained Per-Flow Measurements with Reference Latency Interpolation Myungjin Lee †, Nick Duffield‡, Ramana Rao Kompella†
DaVinci: Dynamically Adaptive Virtual Networks for a Customized Internet Jennifer Rexford Princeton University With Jiayue He, Rui Zhang-Shen, Ying Li,
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.
CS Spring 2011 CS 414 – Multimedia Systems Design Lecture 23 - Multimedia Network Protocols (Layer 3) Klara Nahrstedt Spring 2011.
IP QoS for 3G. A Possible Solution The main focus of this network QoS mechanism is to provide one, real time, service in addition to the normal best effort.
QOS مظفر بگ محمدی دانشگاه ایلام. 2 Why a New Service Model? Best effort clearly insufficient –Some applications need more assurances from the network.
COMT 4291 Communications Protocols and TCP/IP COMT 429.
1 Transport BW Allocation, and Review of Network Routing 11/2/2009.
Korea Advanced Institute of Science and Technology Network Systems Lab. 1 Dual-resource TCP/AQM for processing-constrained networks INFOCOM 2006, Barcelona,
High-speed TCP  FAST TCP: motivation, architecture, algorithms, performance (by Cheng Jin, David X. Wei and Steven H. Low)  Modifying TCP's Congestion.
CS244A Midterm Review Ben Nham Some slides derived from: David Erickson (2007) Paul Tarjan (2007)
TCP with Variance Control for Multihop IEEE Wireless Networks Jiwei Chen, Mario Gerla, Yeng-zhong Lee.
DaVinci: Dynamically Adaptive Virtual Networks for a Customized Internet Jiayue He, Rui Zhang-Shen, Ying Li, Cheng-Yen Lee, Jennifer Rexford, and Mung.
EE 685 presentation Optimization Flow Control, I: Basic Algorithm and Convergence By Steven Low and David Lapsley.
Research Unit in Networking - University of Liège A Distributed Algorithm for Weighted Max-Min Fairness in MPLS Networks Fabian Skivée
June 4, 2003EE384Y1 Demand Based Rate Allocation Arpita Ghosh and James Mammen {arpitag, EE 384Y Project 4 th June, 2003.
TCP: Transmission Control Protocol Part II : Protocol Mechanisms Computer Network System Sirak Kaewjamnong Semester 1st, 2004.
EE 122: Lecture 15 (Quality of Service) Ion Stoica October 25, 2001.
Thoughts on the Evolution of TCP in the Internet (version 2) Sally Floyd ICIR Wednesday Lunch March 17,
Jennifer Rexford Fall 2014 (TTh 3:00-4:20 in CS 105) COS 561: Advanced Computer Networks TCP.
Scalable Laws for Stable Network Congestion Control Fernando Paganini UCLA Electrical Engineering IPAM Workshop, March Collaborators: Steven Low,
H. OhsakiITCom A control theoretical analysis of a window-based flow control mechanism for TCP connections with different propagation delays Hiroyuki.
Murari Sridharan Windows TCP/IP Networking, Microsoft Corp. (Collaborators: Kun Tan, Jingmin Song, MSRA & Qian Zhang, HKUST)
XCP: eXplicit Control Protocol Dina Katabi MIT Lab for Computer Science
1 Traffic Engineering By Kavitha Ganapa. 2 Introduction Traffic engineering is concerned with the issue of performance evaluation and optimization of.
Internet Traffic Engineering Motivation: –The Fish problem, congested links. –Two properties of IP routing Destination based Local optimization TE: optimizing.
Access Link Capacity Monitoring with TFRC Probe Ling-Jyh Chen, Tony Sun, Dan Xu, M. Y. Sanadidi, Mario Gerla Computer Science Department, University of.
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.
Instructor Materials Chapter 6: Quality of Service
Corelite Architecture: Achieving Rated Weight Fairness
Approaches towards congestion control
TCP Congestion Control
TCP Congestion Control
Columbia University in the city of New York
Lecture 19 – TCP Performance
FAST TCP : From Theory to Experiments
On-time Network On-chip
TCP Congestion Control
Review of Internet Protocols Transport Layer
Congestion Michael Freedman COS 461: Computer Networks
Presentation transcript:

End-2-End QoS Internet Presented by: Zvi Rosberg 3 Dec, 2007 Caltech Seminar

What is this talk about  The shortcoming of QoS support in current Internet  A novel holistic Rate Management Protocol  A new scalable QoS guarantee architecture  The theoretical foundation of our architecture  How TCP window flow control may adapt in the presence of our network layer RMP  Another E-2-E prioritized Delay/Loss RMP

Motivation  Shortcoming of current QoS architecture  Beside being immature and requiring horrendous configuration, current QoS also has…  Fundamental inhibitors: 1. Scalability for real QoS guarantee (IntServ and Cisco’s “IntServ over DiffServ”) 2. No bandwidth nor E2E delay guarantee when using a scalable configuration of DiffServ

So what are we doing about it ?  We are implementing a prototype on Network Processors (NPU) addressing the current QoS issues - The architecture is 1. Scalable and has bandwidth, loss and E2E delay guarantee 2. Adaptive - so configuration is minimized 3. Allocates the residual bandwidth fairly  The NPUs execute a new IP layer protocol that router’s should run in the future

The Architecture

The Key Elements of our solution Runs in Edge & Core Routers at IP layer RMP Novel Rate Management Protocol (RMP) for Multi-Service Flows RMP Provides Services to Management functions in the Edge Routers Services

Architectural Components QoS Fair Rate Calculation RMP Link Penalties Gathering Performance Probing Admission Control Scalable Bandwidth Reservation Protocol Classification/Marking at Edge Routers Rate Policing in the Edge Priority Packet Scheduling in Routers Control Plane Data Plane

Theoretical Foundation

Our Theoretical Contribution  Extending Fairness beyond “best-effort” service  Extending the primal-dual iterative distributed algorithm (used by Kelly) for rate allocation with 1. Rate and delay constraints 2. Priority packet scheduling  Revisit TCP flow control when rate is controlled by the network layer  An aside question is: Why priority scheduling?  It improves link utilization – delay-sensitive packets will not have to wait for delay-insensitive packets, so we can have more from the delay-insensitive packets

Fairness with Best-effort  - proportional fairness is equivalent to the solution of: as long as X is convex

Fairness with QoS  A natural way to extend the best effort fairness is to add the QoS requirements to the constraints and …  … optimize on the residual link capacities

 Since X is convex – proportional fairness follows Flow rates of prio 1,2…,m traversing each link maximum loss and delay constraints minimum bandwidth constraints Fairness with QoS (Cont.)

 The delay/loss constraints are NOT EXPLICIT – they are attained by an outer-loop control of Fairness with QoS (Cont.)

Primal-dual iterative distributed algorithm extension  The fair residual rates,, are computed iteratively after a reduction to residual link capacities,, given by  … which is made possible by our scalable reservation protocol  The policed rate of flow is then

The Rate Management Protocol (RMP) In each router output link n and priority m : Total rate of flows from priorities 1,..,m on link n on unreserved link capacity Link capacity reduced by utilization upper bound per priority class m Adaptively set from sources based on RTT and Loss probing Route penalty of flow i

Stability Proof  To prove stability with fixed  We redefine the routing matrix,, to include one virtual link for each priority class  Flows with priority m use all virtual links having priorities m along their original path  The redefined problem is a single class problem equivalent to the priority problem  After this reduction, stability follows by Kelly’s results

Stability Proof (cont.)  To prove stability with adaptive  “Unhappy” flow sources (having excessive delay/losses) signal it in their RMP packets  Congested links decrease the respective  To prove convergence, we allow only to decrease  In practice, convergence is observed also when are also increased when flow sources are “too-happy”

TCP Flow Control - Revisited

TCP Flow Control Revaluation  Once RMP is in place, TCP flow control needs a revaluation  The RMP of the core network will take care of fair rate calculation and congestion avoidance  RMP will also signal end applications about their current target rates, and then…  TCP could be extended beyond “best-effort”  Given rate,, TCP can achieve it with a window update of the form:

Performance Evaluation  We showed that assuming linear scalability, the window flow control converges to a unique stable state under totally asynchronous updates  linear scalability: Total number of bytes queued in each link scales up linearly with the window size  It is an average flow property of the flows crossing a given link, rather a per-flow property  Plausible for large networks  Stability was also verified by simulation  In the fluid model of [Mo & Walrand] used to relate rate and windows, linear scalability is implied

TCP Flow Control Comparison Epoch ISP Network, USA # core links: 74 (37 full-duplex) # flows: 512 # access links: 512 core link capa: 1 Gb/s access link capa: 0.1 Gb/s

Simulation Method  2-way TCP flows using fixed shortest paths  ACKs are either piggybacked or pure (statistically)  RTO is estimated according to RFC 2988 (Jacobson Alg)  Duplicate ACKs are triggered if  All TCP flow controls half their window size upon 3- duplicate ACKs and reduce it to 2 MSS upon RTO  Otherwise - Fast TCP adapts its window sizes according

Simulation Method (cont.)  Simulation time is about 3.5 real operational minutes  In every step - window packets are processed in one batch  First, they are arbitrarily distributed between forward and backward paths  Then, the packets that can “fill” the links are in transit  The rest, are distributed between the bottleneck links in proportion to the bottleneck queueing time  Async operation is modelled by i.i.d Bernoulli r.v's determining which of the flows receive an ACK

TCP Flow Control Comparison Our TCP Flow Control (9 typical flows windows)

TCP Flow Control Comparison Fast TCP Flow Control

TCP Flow Control Comparison TCP Vegas Flow Control

TCP Flow Control Comparison TCP Reno Flow Control (“Sawtooth”)

Comparison Summary Avg Rate Avg RTTAvg WinFairness Dev Max Fair Dev Ours492 P191 ms28 P3%20% Fast479 P231 ms28 P5%25% Vegas449 P248 ms29 P4%44% Reno451 P548 ms59 P12%91%

Flow Control with QoS Support Avg RateAvg RTTAvg Win Priority P50 ms1 P Priority 2224 P56 ms5.12 P Priority 3225 P81 ms7 P  3 x way TCP connections with 3 priorities  Utilization upper bounds: (0.1, 0.75, 1.0)  Avg total fair rate: packets (compared with 492)  Avg Fairness deviation: 5.5%

Simulation with Link Utilization Adaptation  When are adapted based on flow source experienced RTT and Losses (i.e., RTT > RTO), then all QoS requirements are met

Another E2E Delay-Loss Control

Rate Time Derivative in the Fluid Model  clearance time of bits from flows with prio higher/equal p in link l at time t  delay prices for flow i at time t  We study the following prioritized combined Rate-Delay control problem

Delay Time Derivative in the Fluid Model  total rate of flows with priorities less/equal p in link l at time t  The rate control is the gradient search of

Delay Prices Adapting  is learned by the flow source from the RMP packets  … and is adapted if  Adaptation signals must also be disseminated to other relevant sources  …. which is done again with RMP signalling packets

Result Summary  If the routing matrix is full-rank, then  For any e2e delay requirement, there is a unique equilibrium point  The adaptive rate control converges to the stable point from any initial condition Synchronous Fluid Model Time Lag Fluid Model (Rate and Delay effects)  For a single bottleneck case – global stability holds true only if time lag is limited (e.g., ~650 ms)  Emulation – holds true for multiple bottlenecks

Thank You