The idea
Start with the thing that is easy to lose sight of: there is no such object as “the Internet” sitting in a building somewhere. There are millions of separate networks, owned by different people, built from different equipment, running on copper and glass and radio. What makes them behave like one thing is that they have all agreed on the same rules for passing data around.
That agreement is the whole trick. Take it away and you have millions of networks that cannot talk to each other.
So a protocol is not a piece of software. It is a set of rules — how a message is formatted, when it may be sent, what to do when it arrives damaged. Everything else in this unit is a protocol at some layer, so it is worth holding that definition precisely from the start.
Network, internet, Internet
Three words that look like variations on one idea. They are not, and the exam does ask you to separate them.
| Term | What it means | Example |
|---|---|---|
| A network | An interconnection of a group of computers | The machines in one lab, wired to one switch |
| An internet | Any set of networks joined together, lowercase | Two companies linking their office networks |
| The Internet | The specific global system defined by the FNC in 1995 | The one you are reading this on |
The formal definition
On 24 October 1995 the Federal Networking Council passed a resolution defining the term. It sounds like bureaucracy, and in a sense it is — but by 1995 the word was being used commercially and legally with no agreed technical content, and that is a genuine problem when contracts start depending on it.
The FNC said “Internet” refers to the global information system that:
- Is logically linked by a globally unique address space based on the
Internet Protocol (
IP) or its successors - Supports communication using the TCP/IP suite, its successors, or other IP-compatible protocols
- Provides, uses, or makes accessible high-level services layered on that infrastructure
Read those three as a checklist. A private network running IP internally fails the first test, because its addresses are not globally unique. A network that carries IP but offers nothing on top of it fails the third.
Aside
The definition is deliberately written in terms of properties rather than ownership or geography, which is why it has survived thirty years of the thing it describes changing almost completely.
Networks by scale
How it works
LAN, MAN, WAN and access networks
The three main types are separated by geographic scale, and the equipment follows from the scale rather than the other way round.
| Type | Reach | Technologies | Equipment |
|---|---|---|---|
| LAN Local Area | A building or campus | Ethernet, token ring, FDDI, WLAN | Bridges, switches, sometimes routers |
| MAN Metropolitan Area | An entire city | SMDS, ATM, FDDI, wireless mesh and ad hoc | Routers |
| WAN Wide Area | Large geographic areas; long-distance data, voice or video | X.25, Frame Relay, ATM, MPLS | Routers |
Notice that MANs and WANs share their equipment and their role. Both are described as core networks linking multiple networks — their job is joining networks together, not connecting individual hosts. That is why both are router territory while the LAN is switch territory.
The fourth category sits at the edge. An access network connects a private
LAN, WLAN or mobile device to the core network or the global Internet. Base
stations, fibre to the home (FTTH) and xDSL are the examples the unit
gives. If you have a router at home, the link from it to your provider is an
access network.
Where marks get lost
Scale decides the label, not ownership.
A single company can own a LAN in each of its offices, a MAN joining the offices in one city, and a WAN joining its cities. All four belong to one organisation, and they are still four different categories. Answering “LAN, because it is all one company” is the classic way to drop the mark.
SMDS, briefly
The MAN technologies mostly appear as a list to recognise, with one exception
the lecture spells out. SMDS — Switched Multimegabit Data Service —
provides connectionless, high-performance, packet-switched data transport, and
it is built on the IEEE 802.6 Distributed Queue Dual Bus (DQDB) standard.
Worth knowing that “connectionless” and “packet-switched” are doing real work in that sentence; both come back properly in Module 2 when switching is covered.
How it got here
The history is examinable only as a handful of dates and names, so here they are without the padding.
- October 1972 — Kahn runs a large public demonstration of ARPANET at the International Computer Communication Conference. The same year, electronic mail arrives as the first application people actually wanted.
- 1980s — ARPANET splits into a military half (
MILNET) and a civilian half. Other networks appear; the Internet spreads to Europe, Australia, Canada and South America. - 1991 — Tim Berners-Lee creates the World Wide Web at CERN: a simple way to publish information and make it reachable.
- 1992 — The Web becomes publicly available, and growth changes character entirely. URLs start appearing in television advertising.
Robert E. Kahn, with Vinton Cerf, invented TCP and IP. The unit calls Kahn the father of the Internet; if a question asks who is responsible for the protocols at its heart, it is those two names.
In the exam
What actually gets asked from this page:
- Distinguish LAN, MAN and WAN. Give the scale first, then the equipment. A one-line answer that says “LAN is small, WAN is big” will not score; say building or campus, entire city, large geographic areas, and name the equipment.
- Define a protocol. “A set of rules governing all aspects of information communication.” Short, and worth having verbatim.
- What an access network does. Connects a private LAN, WLAN or mobile to the core network or global Internet.
- The FNC definition. Three points. If you can only remember one, remember the globally unique IP-based address space — it is the part that does the defining work.