September 9, 2002 1

Wireless InternetPerformance Research

Carey WilliamsoniCORE Professor

Department of Computer ScienceUniversity of Calgary

September 9, 2002 2

Internet Protocol Stack Application: supporting network

applications and end-user services FTP, SMTP, HTTP, DNS, NTP

Transport: end to end data transfer TCP, UDP

Network: routing of datagrams from source to destination IPv4, IPv6, BGP, RIP, routing protocols

Data Link: hop by hop frames, channel access, flow/error control PPP, Ethernet, IEEE 802.11b

Physical: raw transmission of bits

Application

Transport

Network

Data Link

Physical

001101011...

September 9, 2002 3

The Wireless Web

The emergence and convergence of these technologies enable the “wireless Web” the wireless classroom the wireless workplace the wireless home

My iCORE mandate: design, build, test, and evaluate wireless Web infrastructures

Holy grail: “anything, anytime, anywhere” access to information (when we want it, of course!)

September 9, 2002 4

Research Interests

Wireless Internet Technologies Web Performance Network Traffic Measurement Workload Characterization Traffic Modeling Network Simulation Network Emulation

September 9, 2002 5

Wireless Internet Technologies Mobile devices (e.g., notebooks,

laptops, PDAs, cell phones, wearable computers)

Wireless network access Bluetooth (1 Mbps, up to 3 meters) IEEE 802.11b (11 Mbps, up to 100 meters) IEEE 802.11a (55 Mbps, up to 20 meters)

Operating modes: Infrastructure mode (access point) Ad hoc mode

Classroom area networks (CRAN)

September 9, 2002 6

Example:

Multi-hop “ad hoc” networking

Carey

Gwen

September 9, 2002 7

Example:

Multi-hop “ad hoc” networking

Carey

Gwen

September 9, 2002 8

Example:

Multi-hop “ad hoc” networking

Carey

Gwen

September 9, 2002 9

Example:

Multi-hop “ad hoc” networking

Carey

Gwen

September 9, 2002 10

Example:

Multi-hop “ad hoc” networking

Carey

Gwen

September 9, 2002 11

Example:

Multi-hop “ad hoc” networking

Carey

Gwen

September 9, 2002 12

Example:

Multi-hop “ad hoc” networking

Carey

Gwen

September 9, 2002 13

Example:

Multi-hop “ad hoc” networking

Carey

Gwen

September 9, 2002 14

Example:

Multi-hop “ad hoc” networking

Carey

Gwen

September 9, 2002 15

Example:

Multi-hop “ad hoc” networking

Carey

Gwen

September 9, 2002 16

Web Performance

Explore techniques to improve the performance and scalability of the Web

Examples: Clustered Web servers Load balancing policies Web prefetching strategies Web proxy caching architectures Improvements to HTTP and TCP protocols

September 9, 2002 17

Network Traffic Measurement

Collect and analyze packet-level traces from a live network

September 9, 2002 18

Network Traffic Measurement

Collect and analyze packet-level traces from a live network, using special equipment

101101

September 9, 2002 19

Network Traffic Measurement

Collect and analyze packet-level traces from a live network, using special equipment

Process traces, statistical analysis Diagnose performance problems

(network, protocol, application)101101

September 9, 2002 20

Example Trace0.000000 192.168.1.201 -> 192.168.1.200 60 TCP 4105 80 1315338075 : 1315338075 0 win: 5840 S0.003362 192.168.1.200 -> 192.168.1.201 60 TCP 80 4105 1417888236 : 1417888236 1315338076 win: 5792 SA0.009183 192.168.1.201 -> 192.168.1.200 52 TCP 4105 80 1315338076 : 1315338076 1417888237 win: 5840 A0.010854 192.168.1.201 -> 192.168.1.200 127 TCP 4105 80 1315338076 : 1315338151 1417888237 win: 5840 PA0.014309 192.168.1.200 -> 192.168.1.201 52 TCP 80 4105 1417888237 : 1417888237 1315338151 win: 5792 A0.049848 192.168.1.200 -> 192.168.1.201 1500 TCP 80 4105 1417888237 : 1417889685 1315338151 win: 5792 A0.056902 192.168.1.200 -> 192.168.1.201 1500 TCP 80 4105 1417889685 : 1417891133 1315338151 win: 5792 A0.057284 192.168.1.201 -> 192.168.1.200 52 TCP 4105 80 1315338151 : 1315338151 1417889685 win: 8688 A0.060120 192.168.1.201 -> 192.168.1.200 52 TCP 4105 80 1315338151 : 1315338151 1417891133 win: 11584 A0.068579 192.168.1.200 -> 192.168.1.201 1500 TCP 80 4105 1417891133 : 1417892581 1315338151 win: 5792 PA0.075673 192.168.1.200 -> 192.168.1.201 1500 TCP 80 4105 1417892581 : 1417894029 1315338151 win: 5792 A0.076055 192.168.1.201 -> 192.168.1.200 52 TCP 4105 80 1315338151 : 1315338151 1417892581 win: 14480 A0.083233 192.168.1.200 -> 192.168.1.201 1500 TCP 80 4105 1417894029 : 1417895477 1315338151 win: 5792 A0.096728 192.168.1.200 -> 192.168.1.201 1500 TCP 80 4105 1417896925 : 1417898373 1315338151 win: 5792 A0.103439 192.168.1.200 -> 192.168.1.201 1500 TCP 80 4105 1417898373 : 1417899821 1315338151 win: 5792 A0.103780 192.168.1.201 -> 192.168.1.200 52 TCP 4105 80 1315338151 : 1315338151 1417894029 win: 17376 A0.106534 192.168.1.201 -> 192.168.1.200 52 TCP 4105 80 1315338151 : 1315338151 1417898373 win: 21720 A0.133408 192.168.1.200 -> 192.168.1.201 776 TCP 80 4105 1417904165 : 1417904889 1315338151 win: 5792 FPA0.139200 192.168.1.201 -> 192.168.1.200 52 TCP 4105 80 1315338151 : 1315338151 1417904165 win: 21720 A0.140447 192.168.1.201 -> 192.168.1.200 52 TCP 4105 80 1315338151 : 1315338151 1417904890 win: 21720 FA0.144254 192.168.1.200 -> 192.168.1.201 52 TCP 80 4105 1417904890 : 1417904890 1315338152 win: 5792 A

September 9, 2002 21

Time

Seq

Num

X +

Key: X Data Packet + Ack Packet X

XXXX

XXXX

XX

XX

++

++++

+

++++

++

September 9, 2002 22

Workload Characterization

Try to understand the salient features of network, protocol, application, and user behaviour on the Internet

Example: Web server workloads [Arlitt96] Zipf-like document referencing behaviour Lots of “one-time” referencing of documents Heavy-tailed file size distributions Self-similar network traffic profile

September 9, 2002 23

Traffic Modeling Construct programs and statistical models

that capture the empirically-observed network traffic behaviours

Allows flexible, controlled, repeatable generation of workloads for experiments

Examples: Web client workload model MPEG compressed video model Self-similar Ethernet LAN traffic model WebTraff GUI: Web proxy workload generator

September 9, 2002 24

Network Simulation

Use computer simulation to study the packet-level behaviour of the Internet, its protocols, its applications, and its users

Examples: Improving Web performance over ADSL Understanding the effects of user mobility on

Mobile IP routing and protocol performance Studying the design, scalability, and

performance of Web server and Web proxy caching architectures

September 9, 2002 25

Network Emulation

A hybrid performance evaluation methodology that combines simulation and experimental implementation

A simulator that “talks back” (IP packets)

Examples: Web server benchmarking Wide Area Network (WAN) emulation Web proxy cache performance Distributed applications (Internet games)

September 9, 2002 26

Summary Wireless Internet Performance Lab (UofC) Experimental Laboratory for Internet

Systems and Applications (UofS/UofC,CFI) iCORE Research Team:

Five full-time research staff (Web, perf. eval., simulation, wireless, traffic modeling, network measurement) plus 7 graduate students

Research Collaborations: UofC, UofA, UofS, TRLabs, CS/ECE HP, Telus Mobility, SaskTel, Sun, Nortel…

Do cool, “hands on”, industrially-relevant, applied, practical, and exciting stuff!!