Efficient Gigabit Ethernet Switch Models for Large-Scale Simulation Dong (Kevin) Jin David Nicol Matthew Caesar www.iti.illinois.edu University of Illinois.

Slides:



Advertisements
Similar presentations
1 Network Telecommunication Group University of Pisa - Information Engineering department January Speaker: Raffaello Secchi Authors: Davide Adami.
Advertisements

10/11/2014Anue Systems, Inc. 1 v Telecommunications Industry AssociationTR-30.3/ Lake Buena Vista, FL December.
The Effects of Wide-Area Conditions on WWW Server Performance Erich Nahum, Marcel Rosu, Srini Seshan, Jussara Almeida IBM T.J. Watson Research Center,
Logically Centralized Control Class 2. Types of Networks ISP Networks – Entity only owns the switches – Throughput: 100GB-10TB – Heterogeneous devices:
Analytical Modeling and Evaluation of On- Chip Interconnects Using Network Calculus M. BAkhouya, S. Suboh, J. Gaber, T. El-Ghazawi NOCS 2009, May 10-13,
Chapter 10 Congestion Control in Data Networks1 Congestion Control in Data Networks and Internets COMP5416 Chapter 10.
Priority Scheduling and Buffer Management for ATM Traffic Shaping Authors: Todd Lizambri, Fernando Duran and Shukri Wakid Present: Hongming Wu.
Advanced Computer Networking Congestion Control for High Bandwidth-Delay Product Environments (XCP Algorithm) 1.
XCP: Congestion Control for High Bandwidth-Delay Product Network Dina Katabi, Mark Handley and Charlie Rohrs Presented by Ao-Jan Su.
Submitters: Erez Rokah Erez Goldshide Supervisor: Yossi Kanizo.
Color Aware Switch algorithm implementation The Computer Communication Lab (236340) Spring 2008.
Introduction Future wireless systems will be characterized by their heterogeneity - availability of multiple access systems in the same physical space.
High-Speed Router Design. Content Classes of Routers Components of a Router High-Speed Router Lookup Advances in Switching Fabrics Speeding Up Output.
10 - Network Layer. Network layer r transport segment from sending to receiving host r on sending side encapsulates segments into datagrams r on rcving.
Service Disciplines for Guaranteed Performance Service Hui Zhang, “Service Disciplines for Guaranteed Performance Service in Packet-Switching Networks,”
In-Band Flow Establishment for End-to-End QoS in RDRN Saravanan Radhakrishnan.
Scalability and Accuracy in a Large- Scale Network Emulator Amin Vahdat, Ken Yocum, Kevin Walsh, Priya Mahadevan, Dejan Kostic, Jeff Chase, and David Becker.
ACN: Congestion Control1 Congestion Control and Resource Allocation.
Evaluating System Performance in Gigabit Networks King Fahd University of Petroleum and Minerals (KFUPM) INFORMATION AND COMPUTER SCIENCE DEPARTMENT Dr.
Computer Networking Lecture 17 – Queue Management As usual: Thanks to Srini Seshan and Dave Anderson.
Color Aware Switch algorithm implementation The Computer Communication Lab (236340) Spring 2008.
Congestion Control for High Bandwidth-Delay Product Environments Dina Katabi Mark Handley Charlie Rohrs.
Data Center Traffic and Measurements: Available Bandwidth Estimation Hakim Weatherspoon Assistant Professor, Dept of Computer Science CS 5413: High Performance.
Comparing the Accuracy of Network Simulators for Packet-Level Analysis using a Network Testbed Chaudhry Usman Ali UNB, Fredericton.
Pipelined Two Step Iterative Matching Algorithms for CIOQ Crossbar Switches Deng Pan and Yuanyuan Yang State University of New York, Stony Brook.
Tcipg.org 1 An Alert Buffer Overflow Attack in DNP3 Controlled SCADA Systems Objectives/Problem Investigate a simple but effective attack to block legitimated.
4: Network Layer4b-1 Router Architecture Overview Two key router functions: r run routing algorithms/protocol (RIP, OSPF, BGP) r switching datagrams from.
1 IP routers with memory that runs slower than the line rate Nick McKeown Assistant Professor of Electrical Engineering and Computer Science, Stanford.
A Generalized Processor Sharing Approach to Flow Control in Integrated Services Networks: The Single-Node Case Abhay K. Parekh, Member, IEEE, and Robert.
An Integrated IP Packet Shaper and Scheduler for Edge Routers MSEE Project Presentation Student: Yuqing Deng Advisor: Dr. Belle Wei Spring 2002.
1 Performance Evaluation of Computer Systems and Networks Introduction, Outlines, Class Policy Instructor: A. Ghasemi Many thanks to Dr. Behzad Akbari.
Advance Computer Networking L-5 TCP & Routers Acknowledgments: Lecture slides are from the graduate level Computer Networks course thought by Srinivasan.
TELE202 Lecture 5 Packet switching in WAN 1 Lecturer Dr Z. Huang Overview ¥Last Lectures »C programming »Source: ¥This Lecture »Packet switching in Wide.
An I/O Simulator for Windows Systems Jalil Boukhobza, Claude Timsit 27/10/2004 Versailles Saint Quentin University laboratory.
1. Performance Guarantees Introduction –by asking sources about flow behavior it is possible to construct networks that could guarantee performance for.
Introduction of CRON Lin Xue Feb What is CRON “cron.cct.lsu.edu” testbed project is based on the Emulab system in the University of Utah. Emulab:
Fair Queueing. 2 First-Come-First Served (FIFO) Packets are transmitted in the order of their arrival Advantage: –Very simple to implement Disadvantage:
High-speed TCP  FAST TCP: motivation, architecture, algorithms, performance (by Cheng Jin, David X. Wei and Steven H. Low)  Modifying TCP's Congestion.
Packet Scheduling and Buffer Management Switches S.Keshav: “ An Engineering Approach to Networking”
Dana Butnariu Princeton University EDGE Lab June – September 2011 OPTIMAL SLEEPING IN DATACENTERS Joint work with Professor Mung Chiang, Ioannis Kamitsos,
Bjorn Landfeldt, The University of Sydney 1 NETS3303 Networked Systems.
4/19/20021 TCPSplitter: A Reconfigurable Hardware Based TCP Flow Monitor David V. Schuehler.
CS640: Introduction to Computer Networks Aditya Akella Lecture 20 - Queuing and Basics of QoS.
Hot Interconnects TCP-Splitter: A Reconfigurable Hardware Based TCP/IP Flow Monitor David V. Schuehler
An Efficient Gigabit Ethernet Switch Model for Large-Scale Simulation Dong (Kevin) Jin.
Forwarding.
Modeling and Simulating Time- Sensitive Networking Harri Laine.
An Efficient Gigabit Ethernet Switch Model for Large-Scale Simulation Dong (Kevin) Jin.
An Efficient Gigabit Ethernet Switch Model for Large-Scale Simulation Dong (Kevin) Jin.
Goals The DNP3 protocol is widely used in electrical power systems as a means of communicating observed sensor state information back to a control center.
1 How scalable is the capacity of (electronic) IP routers? Nick McKeown Professor of Electrical Engineering and Computer Science, Stanford University
Queue Scheduling Disciplines
An Efficient Gigabit Ethernet Switch Model for Large-Scale Simulation Dong (Kevin) Jin.
Input buffered switches (1)
1 Lecture 15 Internet resource allocation and QoS Resource Reservation Protocol Integrated Services Differentiated Services.
Ethernet Packet Filtering – Part 2 Øyvind Holmeide 10/28/2014 by.
Team: Aaron Sproul Patrick Hamilton
Empirically Characterizing the Buffer Behaviour of Real Devices
Data Center Networks and Switching and Queueing
Chapter 4: Network Layer
CS 1652 The slides are adapted from the publisher’s material
Advance Computer Networking
ns-2 simulation of TCP + CBR traffic
Design Review.
Advance Computer Networking
COMP/ELEC 429 Introduction to Computer Networks
Uniprocessor scheduling
Introduction to Packet Scheduling
Chapter 4: Network Layer
Introduction to Packet Scheduling
Presentation transcript:

Efficient Gigabit Ethernet Switch Models for Large-Scale Simulation Dong (Kevin) Jin David Nicol Matthew Caesar University of Illinois at Urbana-Champaign

Introduction and Motivation Gigabit Ethernet Widely implemented low-level network technology Backbones of many large scale networks –Power grid control networks Modbus, DNP3, protocols can run on top of Ethernet frames –Data centers –Metro area deployment Frame delay and loss mainly caused by switches 2

Introduction and Motivation Lab Testbed with Real Devices 3 Ethernet Switch Relay SEL 421 Data Aggregator SEL 3351 Control Station Historian

Introduction and Motivation 4

Switch Model Requirements Fast simulation –With embedded real devices test-bed simulation must run faster than real time Accuracy in latency and frame loss prediction that is sufficient for studies of applications and protocols running on the network Straight forward model development for new switches 5

Ethernet Switch Three basic components: buffer, processor and switching fabric Different scheduling policy first come first serve (FCFS), round robin (RR), fair queueing(FQ), weighted round robin (WRR) … Different buffering strategy Shared buffering, input queueing, output queueing, virtual output queueing … 6

Approach 7 1. Perform Experiments on real switches 2. Build Analytical models 3. Build Simulation models in RINSE 4. Evaluate Simulation Speed and Accuracy

Measurement Test-bed Requirement Generate traffic up to line rate with user configured parameters such as frame size, sending rate and inter- frame gap Record frame delays and arrival orderings with microsecond resolution Capture frames at line rate with little loss 8

Measurement 9

10 Single flow Constant Bit Rate Raw Ethernet frames Input rate: 100Mb/s to 1Gb/s

Measurement 11 3COM NetGear Two flows Constant Bit Rate Raw Ethernet frames Each input rate: 100Mb/s to 1Gb/s

Queueing Models 12 NetGear, FCFS 3COM, WRR

Queueing Models 13 3COM Real Queuing Model NetGear Real Queuing Model

Latency-Approximate Model WRR queueing model –One event per frame arrival per switch –One per scheduling round per switch Latency-Approximate WRR model –One event per frame per switch –Upon frame arrival accurate determine if the frame will be dropped estimate latency based on past delay –Not managing the explicit queueing of frames 14

Latency-Approximate Model 15

Lantency-Approximate Model 16 3COM Real Lantency-Approximate WRR Model

Model Evaluation - Speed 17 Q1 Lantency-Approximate model Q2 Detailed queueing model

Model Evaluation - Loss Accuracy 18 Flow 1: 900 Mb/s Flow 2: 100 Mb/s, 200 Mb/s… 1Gb/s 10 million frames per flow Total Input Rate 3COM Switch

Conclusion and Future Work 19 Created a unique test-bed –Generate and capture gigabit Ethernet traffic at line rate with millions of frames –Measure precisely the latency and loss patterns in a switch Different switches has very distinct behavior and require different models Created simple model to reduce execution cost to 60%, almost one event per frame per switch; validated the models with real traffic To understand how switch behaves under realistic bursty traffic

Thanks! ? 20