Tribes and Alchemy Evolution of the Real-Time-Challenged Internet

1
Tribes and Alchemy Evolution of the
Real-Time-Challenged Internet

• Guiding the future of networking by understanding
  the history

P.Asprey Nov 2006
2

• 1950 Many Engineering Tribes
• Telecom, Telemetry, Television, Datacom, Command
  and Control, Power, Sound
• What would the Internet look like if they
• had smoked a peace pipe?

3
Who Said It?

• The idea is idiotic on the face of itthey have
  not the slightest idea of the true problems
  involved.
• Cell phone voice quality has prepared the public
  for VoIP.
• our network has the fewest dropped calls!
• Bell Labs engineers said Packet switching wont
  work.
• IP is so attractive as the packet infrastructure
  because . software applications running over IP
  do not have to be known by the networkNote To
  provide the proper prioritization on a congested
  IP network, it must have some knowledge of the
  applications.

4
Who Said It?

• The idea is idiotic on the face of itthey have
  not the slightest idea of the true problems
  involved.
• -Western Union re Bell Patent
• Cell phone voice quality has prepared the public
  for VoIP.
• -Dr. Treichler, VoIP Industry
• our network has the fewest dropped calls!
• -Cingular advertising slogan
• Bell Labs engineers said Packet switching wont
  work.
• -Roberts of ARPA IPTO
• IP is so attractive as the packet infrastructure
  because . software applications running over IP
  do not have to be known by the networkNote To
  provide the proper prioritization on a congested
  IP network, it must have some knowledge of the
  applications.
• -Cisco white paper on real-time IP

5
Sowing the Seeds - Growing a Communications
Network
1890 - 1930s Hello? Hello? WWI, Aircraft,
Wireless
Telecom Teletype, Switchboards Transmission of
words Transmission of sound Character codes
Baudot Telegraphy Store Forward Messaging
Social and Political Bell Telephone patent 4000
Telcos in 1902 Radio, TV Western Union ATT
Antitrust QoS (Urgent Telegram) Radio
Shack Talkies NY Worlds Fair
20 million phones Public Switched Telephone
NW Television, Telegraphy Sound transmission -
real-time Words transmission -
non-real-time FCC, IEC Standards
6
Era of Big Government Computers 1940-1960

• No high-level policy input from computer
  scientists
• Computing - a new animal
• Requires subtle mix of science, engineering
• Government agencies,military pay for research
• ONR, NBS, DOD, AEC, NASAuncoordinated G-Jobs
• Funded corporations AND were their customers
• Government agencies also built their own
• Military-University-Complex swells
• By 1950 Government spending 20 million per year

7
Sowing the Seeds - Era of Government Computers
1940s Great War, Computing Machines, Cold War
Telecom Traveling Wave Tube Klystron, Magnetron
Microwave, FDM
Computing ENIAC Whirlwind Computer Stored
Program Primitive Transistors Real-time
Computing
Social and Political Radar Manhattan Project/Los
Alamos US gets The Bomb Soviets get The
Bomb Balance of Power Military-University-Complex
Artificial Intelligence MIT AI Lab
30 Million Telephones Fear! Money! Baby
Boomers Basic Computing and Transmission
Technologies
8
Sowing the Seeds - Era of Government Computers
1950s USA Good Guys, Domino Theory, Space Race
Computing Miniaturization Cheap
Transistors Magnetic memory Magnet disk storage
COBOL, FORTRAN LISP, ALGOL
Social and Political Military work is cool ICBM
Detection USSR Sputnik NASA, ARPA Admiral
Grace Hopper Double Helix
Telecom Nationwide Dialing Electronic
Switching Modems Digital Multiplex T-1
50 Million Telephones Stable Digital
Hardware Large, Expensive Computers Compilers,
Languages Programming is Interesting
9
Growth - Commercialization of Computers
1960s Military Industrial Complex, Cool
Scientists, Star Trek
Social and Political Vietnam Anti-War
Protests Drug Culture,Genes, Memes NASA - Fault
Tolerance NSF Computing Education ARPA
Visionaries CompuServe sells timesharing Kleinroc
k - Packet Switching Theory Mad scientist Dr.
No,
Telecom Dataphone modem Telstar - Voice,
TV 24 channel TDM Digital switching Digital
signal processing Digital image processing
Computing MULTICS, UNIX High-speed
Data/Fax Integrated Circuits IMP NW
Interface Redundancy Dual Core Computing
90 Million Telephones Critical research mass,
Stable funding ARPA Graduates Timesharing,Busines
s Computing
Bond, James Bond.
10
Golden Age of US Research Policy1960-1970

• Eisenhower warning rise of the Military
  Industrial Complex
• Scientists in high places - Eisenhower/JFK
• Stable funding, long-term projects, critical mass
• ARPA /IPTO (Information Processing Technology
  Office
• under civilian control - JFK
• Commercial applications.
• Banking, airline reservations
• Government still paying for internal/external RD
• Telecom research - Telcos and Military
  Communications

11
1960s Scientists in Charge Creating The Vision
and Funding the Research

• New ARPA IPTO - Information Processing Technology
  Office
• Crazy is OK Mechanical elephant1 crt, many
  computers
• ARPA Diversity Psychologists and Engineers (no
  women)
• Licklider - Man-Computer Symbiosis, Galactic
  Network
• Sutherland - Computer graphics, Snake Robot
• Taylor - Interactive Networks, 3 Terminals
  Problem
• Roberts - Cooperative Networks
• Baran - Secure Packetized Voice (Military)
• NSFNet backbone for research only
• Internet NSFNet merged with Arpanet

12
Crosstalk? PSTN vs Packet SwitchingReal-Time
Digital Divide

• US already had an extensive communications
  network experience
• Public switched telephone network (PSTN)
• Telemetry networks real-time, store-and-forward
• Data networks (Teletype) and Real-time networks
  (Voice, Radio, Broadcast)
• Computer Scientists vs Telecom Engineers
• Different language, journals, conferences,
  terminology
• Telecom - improve quality, reliability of voice
  NW
• Datacom engineers Voice over IP seems to work

13

• Compares call quality scores for different
  service providers
• Colors represent call quality MOS (Mean Opinion
  Score) rating
• Most VoIP calls currently have unacceptable
  quality rating
• Source
• Ellacoya Networks

• Explosion of VoIP traffic in last 9 months
• Over 250 different VoIP providers
• Decrease in average VoIP quality reported in last
  3 months

14
Developing the Networks
1970s Phone Phreaks, PC Revolution, Mad Computers
Social and Political ARPA to DARPA Xerox PARC
(Taylor) CS without EE degrees Vinton Cerf,
NCP Wozniaks blue box Entrepreneurs Jobs,
Apple Gates, Microsoft
Telecom International Dialing Digital CO
Switch International Email Telephone hacking
Computing, Networking UNIX, C PASCAL,
Smalltalk Mini, Microcomputers Dial-up NW, Packet
Radio TCP/IP
Standards, OSI Model, Structured Programming PCs,
email, Basic Internet, Demystification of VLSI
HAL 9000 2001, A Space Odyssey
15
1970s ARPA Gets a D Defense - More Development,
Less Research

• Japanese computer threat (like the automobile)
• US Semiconductors business drops 75 to 40
• Government-supported research military
  relevance
• General skepticism about the role of science
• DOD funding for mathematics, CS two-decade low
  in 1975
• 1986 dropoff in engineering students (1
  generation)
• ARPA director - industrial background
• Applications with short time horizons, months
  instead of years

16
Developing the Network - Cyberspace
1980s Desktop Computing, Email, Divestiture,
Netheads
Social and Political Berlin Wall Falls DOD in
decline Drives less RD NSF Adds to
Research Dialup Computer Services (ruined Bell
traffic models) Netheads vs Bellheads Powerful
Desktops SUN High-speed graphics, Ethernet, VLSI
Telecom Synchronous Optical Network TAT-8 ATM/Broa
dband Integrated Services Digital Network
Computing Electronic Bulletin Boards TCP/IP .gov
.com .edu SEMATECH public/private 40 homes
w/computers Accidental virus Internet Worm
Huge Base of Users Domain Names, Optical NW Less
basic research, Anti-virus Software
17
Color Optics and Telecome Overinvestment
"Surfing the net" -Vint Cerf
1990s Color Optics, World Wide Web, Internet
Bubble
Social and Political http.//www. WorldWideWeb
browser Berners-Lee, CERN Scramble to support
research Secure Internet Transactions E-commerce V
oIP (Toll bypass) 3rd Generation Partnership
Project Net traffic Doubles every 6 month Yahoo
Networking data usersphone users users WWW.,
browser, Secure Socket Layer Component software
JAVA, Color optics VoIP standards, IP PBXs
Overbuilt Backbone hides QoS Failings, Internet
Bubble Browsers E-Commerce, International email
18
Dot.Bombing and Recovering the Internet
2000s Broadband, World Wide Wait, Spam, What
QoS?
Social and Political Peering VoIP
difficulties The Google Real Y2K problem is
.com Bomb Social Networking Viruses and SPAM IP
Multimedia Subsystem (IMS) Next Generation
Network (NGN) IPv6 supports ATM, MPLS ideas Flow
Routing Net Neutrality concept
Internet Cellular phones gt wireline phones IP
related revenue almost 3 GWP Carrier Grade VoIP
not a reality Flow routing ideas
Google is a Verb Problems with Real-Time IP,
Peering, Digital Divide, Network Neutrality
Issues Social Networking
19
Internet and Beyond IP Two tin cans and a
string -anon

• TCP/IPv4! Phenomenal results after two decades.
• Must re-examine technological assumptions
• cheap memory and processors
• huge bandwidth
• Must define next generation requirements
• crazy is OKe.g. real-time applications work
• Must create the vision and fund the research
• US RD flat at 2.6 percent of GDP for FOUR
  decades
• Federal share dropped from 2/3 to 1/4 of that
• Must NOT entrust research to corporations!

20
Alchemy! Innovation and Wealth The greatest
legal creation of wealth in history-John Doerr,
VC

• During the Cold War, millions of taxpayer dollars
  were funneled into basic research in computing
• That Money emerged 40 years later as private
  wealth
• Start-ups and IPOs
• VCs and Investment Bankers
• Is that money being re-circulated into US basic
  research?
• Lear Jets and vacation homes
• The Trickle Out theory
• Are innovators rewarded proportionately?
• More MBAs than Engineering degrees
• Quality and reliability - casualties of this gold
  rush?

21
Real-time Voice/Video and Data Traffic How
Different Are They? Fundamentally.

• Voice or Video Traffic
• Real-Time (human speech)
• Continuous, full duplex
• Tolerant of errors, lost packets
• Intolerant of transmission delays
• Lost packets cant be retransmitted
• Data Traffic
• Not Real-Time (email, web browsing)
• Bursty
• Intolerant of errors, lost packets must be resent
• Tolerant of delays

22
Real-Time and IP Layer 8 The human eye, ear,
brain

• IP Philosophy - Let the higher-level protocols
  recover from lost packets
• 3 million years of evolution, no upgrades likely
  soon!
• different protocols, different ISP delays,
  firewalls,..
• ATM Forum Just give us a bit - Netheads
• IP by-passes QoS potential of ATM
• Solving a different problem (data)
• MPLS - ATM without cells
• RTP - Real-Time Packet protocol
• Sequence and relative timestamp
• Anti-tromboning
• Now you KNOW you have jitter!

23
Tragedy of the CommonsNet Neutrality Every
packet for itself!

• Peer-2-peer networks, Gamers, Spammers
• Google rankings ploys
• Myspace media madness,
• Broadcast TV, Video on demand
• Reliable User Data Protocol (RUDP)
• Sends multiple copies of the same packet
• Receiver discards redundant packets
• ONE of the packets will make the journey!
• Excellent solution for congestion Add packets!

24
Strong Rise in Symmetric Peer-to-peer Applications
P2P Symmetrical Usage

• More than ½ of all Internet traffic is
  peer-to-peer file sharing
• Soon robotic application traffic will dominate
  networks
• P2P traffic is often symmetric (same amount
  upstream and downstream)
• thwarts traffic models and architectures

25
The Perfect Network

• Adjust network to user, not user to network
• Self-healing when a node is lost
• Self-provisioning when a node is added
• Self-balancing network, minimal traffic
  engineering!
• Dont let me start unless I can finish
• Closed-loop applications - remote medicine
• Priorities based on USER choices
• Real-time traffic accomodated
• Secure, control spam, denial of service, etc.

26
Who? The network is the computer. - Scott
McNealy

• Desktop computing has grown too complex, ugly
• Backup, security, compatibility, cookies,
  browsers, etc.
• Frustrated users aboundnerd market saturated
• ASP (Google-type) Calendar - other office
  applications?
• Are Traditional carriers are the most logical
  providers of the
• new network?
• Have done dialtone, videotone, and wirelesstone
  for a long time
• More reliable, predictable, and credible than
  computer industry
• Ubiquity Web-site tone, you-name-it tone to all
  of us
• Know how to bill millions of people for billions
  of transactions
• Have the infrastructure and field service, the
  man in the van

27
Now What?

• Fund science education (eliminate AP Classes?)
• Fund taxpayer-owned basic research
• Civilians in charge, NASA, IPTO as models
• No corporate secrecy or backwards compatibility
  requirements
• No company can bury or force direction on
  research
• Research is NOT a Manhattan Project. Start
  soon!!!
• Do we regulate web-tone for the Better Good?
• E.g. ATT Long-distance service subsidies for
  remote services
• Need reliable infrastructure, not peering
  problems
• Go beyond entrenched IP Players - create diverse
  sets of network designers and dreamers (women
  too!)
• . Smoke a Peace Pipe and share some knowledge!

28
Final Who said it?
The term "cyberspace" has past its sell-by date
The problem is that everything has become an
aspect of, well, cyberspace The internet feels
less like an alternate world that we 'go to' and
more like just another layer of life."
29
Final Who said it?

• The term "cyberspace" has past its sell-by date
  The problem is that everything has become an
  aspect of, well, cyberspace The internet feels
  less like an alternate world that we 'go to' and
  more like just another layer of life."

William Gibson (Coined the term cyberspace in
his 1984 Neuromanser sci-fi novel.)