1. Internet History
Hobbes Internet Timeline

2. People, ideas, projects, and applications
The history of the Internet is the story of people and their ideas, and the projects they worked on.
It is not the story of any one person or project
It is not the story of any single application, whether electronic mail, remote login, file transfer, the web, video, or machine-to-machine communication
Or even of ICANN or the IANA
However, many people and projects made important contributions.
This is an attempt at the story

3. The original concept: Survivable networks

4. 1961: Leonard Kleinrock wrote on packet switching concepts His logic:
Subsequently convinced Larry Roberts to look at packet switching as an alternative to circuit switching
His logic:
If the strength of a chain is that of its weakest link,
Then the strength of a network is the strength of its last surviving path

5. 1962: J.C.R. Licklider’s (MIT) "Galactic Network" concept
He envisioned a globally interconnected set of computers through which everyone could quickly access data and programs from any site.
“Wouldn’t it be cool if one could walk up to a computer
and find the knowledge of the universe at our fingertips?

6. Also 1962:
Paul Baran at RAND independently suggested that a packet-based network might be a useful way to build a survivable voice network,
Baran was the one that suggested we call them “packets”. We do.

7. 1965:
Roberts and Merrill connected two computers – and found circuit switched analog telephone inadequate for their purposes

8. 1967:
Larry Roberts, DARPA program manager, started research on the concept of packet networking

9. 1968:
Roberts issued an RFP for research, which was soon awarded to BBN
The network BBN developed, called the ARPANET, had its first node with UCLA and Len Kleinrock
The second node was at SRI
The first packet to cross the network went from UCLA to SRI – an attempt to log into a computer

10. 1969:
The RFC Series was originated as a way to share notes among researchers – Steve Crocker
The notes were called “Requests for Comments” in an attempt to downplay their importance.
Later, vendors would joke that they were “Requirements for Compliance”, e.g., specifications
Many were in fact white papers, thoughts about what might be – and what might in the end not be - blind alleys…
The ruminations of a a far-flung distributed research laboratory on topics they were just working out

11. Context: fledgling communications
Synchronous “start/stop” protocol patented 1966
IBM 3780/2780 Binary Synchronous Communication
IBM 3270 Binary Synchronous Communication
HDLC/SDLC development

12. Sliding Windows
Early transmission protocols had crude ways, if any, to identify and recover from errors
Master/Slave
Poll/Call
Half Duplex (two way alternate)
SDLC/HDLC made a giant step forward
Full Duplex (two way simultaneous)
Using a sliding window, it could keep data being transmitted (or retransmitted) in one direction while being acknowledged in the other, and in LAPB, data continuously in transit in both directions.

13. 1970: NCP prototyping and deployment on the ARPANET
ALOHA, a satellite network based on random or semi- random transmission

14. 1972 Louis Pouzin invented the datagram:
A packet that contains all necessary state within itself and so depends on no external network state
Operating on a best-effort basis – it may be lost, duplicated, or reordered in flight

15. Also 1972: Robert Kahn demonstrated the fledgling ARPANET at ICCC
The application that he demonstrated:
Ray Tomlinson’s electronic mail
Original telnet (remote login) specification (RFC 318)
John Postel

16. Kahn’s Ground Rules
Each distinct network would have to stand on its own:
No internal changes could be required to any such network to connect it to the Internet.
Communications would be on a best effort basis.
If a packet didn't make it to the final destination, it would shortly be retransmitted from the source.
Black boxes (IMPs) would be used to connect the networks;
These would later be called gateways and routers.
Gateways retained no per-flow state, thereby keeping them simple and avoiding complicated adaptation and recovery from various failure modes.
There would be no global operational control.
Sites were by definition autonomous. The only protocols they had to implement were IP and ICMP, and maybe TCP and UDP

17. 1973 and on: Kahn/Cerf and TCP
The original protocol combined the services of what we today call “TCP” and “IP”.
It worked well for file exchange and remote access to time- sharing systems,
But not some other applications, for which application control was more important
The update process:
Separated TCP from IPv4 (RFC 791, 792, 793)
Added UDP for applications that needed it
Electronic mail, remote login, and file exchange remained the “killer applications” for two decades or more

18. Starting in 1973:
Development of concepts for a random local area network (one that didn’t cross a legal boundary) by Bob Metcalf
That was experimented with at Xerox PARC (PupNet)
Resulted in the DEC, Intel, and Xerox specification for the Ethernet, 1981

19. 1974:
BBN Telenet: first commercial packet data service

20. Early 1980’s January 1, 1983: 1984: Cut-over from NCP to TCP/IPv4
Deployment of the Domain name System
Replaced centrally-managed “hostfile” with a distributed and recursive system for naming
Names originally translated simply to IP addresses or lists of names of mail servers

21. And then Al Gore… …Got Money
Circa 1984, the junior senator from Tennessee started discussing his ideas of an “Information Superhighway”.
He had five NSF-sponsored supercomputer centers, and he wanted to connect universities to them
NSF-funded IP networking experiments:
CSNET: an X.25 network
USAN: a wide area Ethernet network over satellite
56 KBPS NSFNET
Other networks:
NASA Science Internet
CYCLADES
CERN networks
BITNET …
…Got Money

22. Regional networks
The NSFNET grew dramatically, as universities bought routers and connected first to it and then each other
Changed successively from 56 KBPS to T-1 to T-3
States or other agencies built regional networks that connected to it: NEARNet, BARRNET, PSINET, NYSERNET, and many others

23. Commercialization of the backbone
1990:
the ARPANET Core, having become irrelevant after 22 years of operation, was shut down
Formation of the IANA at USC/ISI
Recognition that IPv4 would eventually run out of address space – starting work on CIDR and IP Next Generation
Initial formation of Regional Internet Registries
ARIN, RIPE, APNIC
LACNIC and AfriNIC came later
Succeeding years:
Many regional networks in time became commercial networks.
And in turn replaced the NSFNET as the core network

24. Competing network technologies
These existed in the 1980’s, interconnecting LANs (subnet and host)
These existed and could interconnect companies
SNA
DECNET
AppleTalk
XNS Internet Transport
XNS
3COM
Ungermann-Bass
Banyan Vines
Novell Netware
MIT Chaosnet
Sytek NetBIOS
others
Connection-Oriented and Connectionless OSI
Address (NSAP) identifies network, subnet, and host
IPv4
Originally identified network, subnet, and host
With CIDR (1992), could aggregate networks to identify service providers

25. Changing applications
Every 3-5 years, the Internet fundamentally changes in the payload it carries
: SMTP, FTP, Network News, telnet
1992: World Wide Web, multicast, experimental voice/video
1995: WWW with multiple sessions in parallel, commercial Voice on IP
2000: Peer to Peer file sharing in various forms
2003: Web 2.0 applications like MySpace, Facebook, BitTorrent File Sharing
2008: Cyberlockers replacing file sharing
1990-present: Rise of video in various forms
Lately: Map/Reduce and Hadoop – data center distributed applications
Next…
Not that old payloads go away: we add new and sometimes dominant payloads in addition to the old
On the commercial backbone, video is becoming dominant, primarily from Content Providers that colocate with an ISP’s POPs or data centers
In private networks (Smart Grid, Health Care, public and private safety, industrial automation) we see distributed telemetry and distributed control.
Changing workloads - a normal thing

26. The Internet in 2013 The structure Kahn’s principles:
Perhaps 10,000 networks offering communication or content services to others
Roughly another 50,000 networks interconnected using BGP routing
Billions of users world-wide
The primary service of the network is connectivity.
The Internet thrives on innovation, and fosters it.
Each distinct network is autonomous
Communications is on a best effort basis.
Routers are used to connect the networks;
There is no global operational control.

27. Internet History
Hobbes Internet Timeline