When we discuss copper, fiber and wireless, we focus on the media, often without regard to the networks for which the media are supposed to provide connections. To better understand the role of the media and the selection of the best choice, it helps to understand networks.
What is a network? It is basically a group of interconnected devices that can communicate with each other and maybe devices in other networks, too. Media connections allow for sending messages, sharing files and working in unison. The devices need a way to connect to each other, either over cable (copper or fiber) or wireless. This connection is called the “physical layer” in standards.
All the devices in a network share the network connections. Sharing may be accomplished by connecting devices on separate paths, sequentially connecting devices using time-division multiplexing or simultaneously sharing the media using frequency multiplexing (such as CATV) and wavelength-division multiplexing in fiber optics.
Since devices are sharing the network, they need unique identification and a unique address to send communication from and to. They also need some rules for gaining access to the network without interfering with other devices’ messages. Those rules are called the network protocol.
The telephone network was the first real network. In a plain old telephone service (POTS), one phone had to be physically connected to another to have a conversation, creating what is called a star network. Signals were transmitted over a pair of wires using a current loop. In the early days, switching was done by operators patching wires at a switchboard, then by mechanical switches and finally by electronic switches. Switching was done along preprogrammed routes to allocate available bandwidth and have alternative routes if one was busy.
With the advent of digital phone systems, speeds were high enough that, on top of physical switching, one could share a connection among many users by sequentially transmitting digital signals for each conversation. This is called “time-division multiplexing” (TDM) to introduce one of many three-letter acronyms (TLAs) that are used in telecommunications. A digitized phone call required only 64 kilobits per second. An early T1 digital phone line at 1.544 megabits per second could multiplex 24 calls. I will leave it as an exercise for the reader to figure how many calls one of today’s 10–100-gigabits-per-second phone links could carry.
The combination of physical switching and TDM was called time-space-time (TST) switching, but TST was inefficient. If there was silence on a phone call, the system still transmitted bits, even though there was no content, tying up otherwise useful bandwidth.
It was not long after computers became available that users wanted them to communicate with multiple peripherals and other computers. The first networks used simple switching protocols like the phone system.
In the mid-1970s, Xerox Palo Alto Research Labs invented a new type of network: Ethernet. In Ethernet, connected devices shared a coaxial cable data bus: the “ether.” Rather than switch onto the network, a connected device would simply listen for traffic. If it heard none, it would transmit its message, called a data packet. If two devices did so simultaneously, causing a collision, they would stop, wait a random time and try again. Like everything techie, this has a catchy acronym: CSMA/CD for “carrier sense multiple access with collision detection.”
The problem with the Ethernet protocol was traffic. More connected devices means more collisions, which means wasted bandwidth. With enough devices, one could not guarantee a device would ever get to send its data, a big problem for systems where some devices (e.g., alarms) need priority. The solution is for the network to work faster. Bandwidth, as we shall see, solves most network problems.
IBM took a different approach. Its network connected devices in a ring on shielded twisted-pair cable. Each device would be granted access to the network by receiving a “token”—a data packet that gave permission to transmit data. Once finished transmitting, the device would release the token to the next user, guaranteeing access to the network. Logically, the network was called IBM Token Ring (TR). The token guaranteed access, so TR was called a deterministic network.
So now we have three examples of network protocol: TST, CSMA/CD and TR. All are designed to allow users to share the available bandwidth of the network. We also have three types of cable: unshielded and shielded twisted-pair and coaxial.
During the 1980s, rapid growth of personal computer networks plus two major technical developments in cabling—balanced transmission on unshielded twisted-pair copper cable and fiber optics—dramatically changed networking. I’ll discuss that next month.