Reference Series Table of Contents For This Issue

How Computers Work, Part I
August 2001• Vol.5 Issue 3
Page(s) 132-137 in print issue

Get Connected With Networks
Hooking PCs Together Increases Their Usefulness
In the span of a few short years, the Internet has made an impressive mark on society. School children now gaze at treasures in distant museums right from their classroom computers while home users idly watch real-time video from rooftop cameras in cities a continent away.

At the heart of this change is the networked computer. Few PC users seem to really understand just how our globe-spanning networks function. Yet it is not too hard to learn if you take things a step at a time, beginning with some of the basic building blocks of networking.

 Linking Computers. A computer network is a collection of two or more computers and associated devices linked together using some form of communications equipment. They are linked together so they can share software, hardware, and information. A group of network computers physically linked at the same location forms a LAN (local-area network). A group of LANs linked over a large area, such as across a state, forms a WAN (wide-area network).

For the single home PC connected only with the Internet, the most obvious piece of communications equipment is the computer modem. Yet in settings where larger numbers of computers exist, such as banks, public utilities, and universities, PCs are linked together using network adapter cards, also known as NICs (network interface cards; pronounced nick). These adapter cards transfer data at rates much higher than those possible with conventional modems.

A sophisticated wiring and switching system connecting the network adapters helps route data from one computer to the next. The type of cabling you use for networks varies from simple telephone wire to coaxial cable to fiber-optic cable. In addition to physical connections, networks use a number of wireless communication schemes, including infrared and RF (radio frequency) transmission.

 It’s In The Cards. Network devices must have some form of unique identification in order for them to tell each other apart. This is the reason why NICs come with an unchangeable 12-digit hexadecimal address.

This address is called the physical address, or the “burned-in” address. You may see the physical address referred to as a card’s “BIA.” A hexadecimal numbering system is a base-16 numbering system that uses combinations of 16 digits to represent all numerical values. In addition to the 10 digits used in the familiar decimal system (0 to 9), hexadecimal numbering uses the letters A through F.

Because we don’t usually think of long numbers as names, we can opt to give a different name to each machine on the network. These names are known as machine names, or logical names. Each machine name has to be unique, as well. For Microsoft networks, these names can be up to 15 characters long.

Cisco Systems serves as one of a variety of companies that specialize in answering networking needs.
Once a NIC is in a PC, the OS(operating system) controls the NIC’s electronic components through software drivers. The card manufacturer usually writes this software. While the software drivers link the card to the OS, network protocols govern how computers communicate with each other.

In Microsoft OSes, the NetBEUI (NetBIOS extended user interface) protocol is used for small LANs. NetBEUI supports network communications. However, in larger networks, the TCP/IP (Transmission Control Protocol/ Internet Protocol) governs the communication lines in many LANs.

 Playing By The Rules. Network protocols must be able to work with a variety of OSes and network hardware. The ISO (International Organization for Standardization) developed the OSI (open system interconnection) reference model to standardize how computers exchange data across a network. In this model, the network data moves up and down through seven different layers.

At each layer, the network examines and processes the data according to the abilities of that particular layer. The network performs more and more complex tasks at each layer.

Physical layer. In the physical layer, the first layer, all network data must enter and leave the PC. The physical layer represents the hardware connections where data moves in and out as a stream of binary data (1s and 0s). This layer includes network cabling, the connectors (which link the cabling to the NIC), and the NIC.

Data link layer. Data from the physical layer moves to the data link layer where the network arranges the data into data frames, packets of information transmitted as a single unit. (Depending on the type of network, these data frames may also be called network packets.) Each frame contains control information as well as the starting point and ending point of the unique information within the frame. This unique information is often referred to as the data frame’s message. The NIC’s software drivers operate in the data link layer.

Network layer. The network layer acts much like the network’s own traffic cop, examining each data frame’s control information to determine the frame’s route to its destination. To do this, it has to be able to compare the logical network name in the control data to the physical address of the NIC. This layer is a very busy place in which frames are sent to their correct destinations.

Transport layer. After leaving the network layer, data frames enter the transport layer. Here, the network examines data frames for errors and places them in the correct order without duplicate or missing frames. This layer is the train yard where the engineer puts all the boxcars in their proper order before the freight train leaves for its destination. This layer ensures complete data transmission.

Session layer. At the session layer, the network examines data frames to make sure they start, continue, and end a logical link between two computers. Using information from the data frames, the session layer determines which computer transmits or receives data and for how long.

Presentation layer. In the next stage, the network passes data frames to the presentation layer. At this layer, the network examines the message data and formats it accordingly. This includes character and number formatting for correct use at the next layer. The presentation layer is also responsible for data security and provides data encryption services.

Application layer. The final layer data passes through is the application layer. Here, the data frames are available to a user’s applications, such as an e-mail program, a FTP(File Transfer Protocol) client, or an Internet browser.

It’s important to remember that all network devices use the OSI reference model. This allows a wide array of OSes from Windows to Macintosh and Unix to all use the same network protocols, such as TCP/IP, to share data. Network protocols generally do their business in the network, transport, and session layers.

The rules laid down by the OSI reference model tell you that network protocols are responsible for transmitting and receiving network data and assembling the data into a usable form for your computer’s programs.

 The Lay Of The LAN. The design of a particular type of network is the network’s topology, and wouldn’t you know it, another organization is responsible for network topology standards. In February 1980, the IEEE (Institute of Electrical and Electronic Engineers; set up the 802 Project to develop LAN standards. The 802 Project devised standards for NICs and a variety of different network components. For instance, you will find information about Ethernet networks in Project 802.3, Token Ring networks in Project 802.5, and network security in Project 802.10.

Ethernet has proven to be, by far, the most prevalent network topology. First developed in 1976 at the Xerox Palo Alto Research Center (which is famous for many computer advances), Ethernet is a baseband network. All its network nodes have equal rights to both transmit and receive data. At this stage in time, Ethernet equipment can operate at either 10Mbps (megabits per second) or 100Mbps. Technicians commonly refer to the 100Mbps Ethernet as the fast Ethernet.

All Ethernet equipment uses the CSMA/CD (collision sense multiple access/collision detection) method. In this scheme, network adapters “listen” to the network to hear if any other adapter is currently transmitting. If no other adapter is transmitting, then the listening network adapter is free to transmit.

To best understand how this works, let’s imagine a very large family is having a Thanksgiving dinner together. Everyone is free to talk, and often people interrupt each other. When this happens, the pace of the family conversation slows down as the talkers pause to allow the other person to speak next.

This is exactly how CSMA/CD works for Ethernet networks. By now, you may suspect that if a large number of Ethernet adapters try to transmit at the same time, these adapters will also pause to allow other adapters to transmit. Actually, they all don’t pause the same amount of time or they’d never be able to transmit their data. Nonetheless, as the Ethernet network gets busier, the average speed of the LAN gets slower. It’s best to think of 10Mbps and 100Mbps as peak speeds for these slow and fast Ethernet networks, rather than as constant speeds.

IEEE (Institute of Electrical and Electronic Engineers; read I-triple-E), as a mission, promotes the engineering process of creating, developing, integrating, and applying knowledge of electrical and information technologies for humanity and the profession.

 Ethernet Flavors. In the current state of networking, there are three physical versions of Ethernet: 10Base-2, 10Base-5, and 10Base-T. The 10 stands for 10Mbps. The term “Base” refers to the type of network. Ethernet, as you will remember, is a baseband network. And finally, the last digit identifies what type of connecting cable the network uses.

In 10Base-5 LANs, a half-inch (in diameter) coaxial cable connects network adapters from up to 500 meters away. It is also known as thicknet Ethernet because of the diameter of the cable. The 500 meter distance gives us the “5” in 10Base-5. When thicknet Ethernet acts in this capacity, it serves as the network “backbone” because it links several distant LANs as part of a larger network.

The 10Base-2 LAN uses a thinner type of coaxial cable that is about a quarter-inch in diameter. It should not come as too much of a surprise that it’s also known as thinnet Ethernet. The smaller size of cable makes it more flexible and easier to thread behind furniture than the half-inch thicknet cable. The major trade-off is the maximum distance for the 10Base-2 LAN is only about 185 meters. This distance gives us the “2” in 10Base-2.

Both 10Base-2 and 10Base-5 are bus-style networks because the network cables are connected to each adapter from two directions, making a daisy chain. (A daisy chain is a set of hardware components that connect to each other in a series.) The end network adapters have to be terminated on one side with a termination resistor, which tells the network hardware where the ends of the LAN lie.

If a cable breaks or it turns out that one of the termination resistors is not in place, all network traffic on the affected LAN ceases. Tracking down cable faults or bad adapters on a bus-style network is, in spite of its simple design, time consuming and tricky.

 10Base-T. Fortunately, Ethernet equipment designers came up with a better Ethernet flavor: 10Base-T. This type of card uses eight-conductor, telephone-style cables known as twisted-pair wire. This is the “T” in 10Base-T. The maximum distance between two points on a 10Base-T network is about 100 meters.

In spite of the number of conductors, this is much smaller cabling that is easier to thread through walls and behind furniture. Because it is standard telephone wire, trained wiring installation crews may only be as far away as your local phone company.

The 10Base-T network adapters are linked together using a network hub. Electrically, this connects all the network nodes together in a star pattern. This means that workstations at each point of the star can talk to any other workstation without having to pass through another workstation. This makes the 10Base-T network a better performer than the bus-style 10Base-5 and 10Base-2 networks. Also, a failure at any point of the star will not bring down the entire network.

Generally, you can connect 10Base-T workstations using standard UTP(unshielded twisted-pair) cable, common in telephone installation. However, this wire can accidentally pick up spurious electrical signals from RFI(radio frequency interference) produced by electric motors or noisy florescent light.

UTP cable is rated according to its ability to pass electrical signals at various frequencies. The EIA/TIA (Electronic Industries Association and the Telecommunications Industries Association) classify the speeds of twisted-pair cable. Cable that is certified to carry data communications at 10Mbps is rated as category 3 cable or CAT 3 (pronounced “cat three”).

You will need a more reliable type of twisted-pair cable to reach the 100Mbps speeds of fast Ethernet. Called CAT 5 cable, this type of UTP cable has more twists in it than CAT 3 cable. The extra twists help reduce the amount of electrical noise picked up by the cable. After CAT 5 cable is installed, special test equipment is used to certify the cable as acceptable for higher-speed network traffic.

In high-interference areas, STP(shielded twisted-pair wire) cable works best. This type of cable includes a metallic foil sheath that surrounds the wire pairs and conducts induced electrical signals away from the cable wires. Its nature also allows it to carry signals for longer distances. Naturally, it is more expensive than UTP cable.

 Linking LAN Segments. If you are going to network computers in an office building, your proposed LAN may span several floors or cabling may need to go around a load-bearing wall. If you are building a 10Base-T network, you can extend your LAN past the 100-meter cabling limit.

When this happens, network technicians use a repeater to extend the maximum length. A repeater serves as a signal amplifier, taking a weakened signal and restoring it back to its original condition. Network technicians refer to this as data regeneration.

Certain networks, such as a 10Base-2 or 10Base-5 network, can also use repeaters to extend the number of computers in that network. In a 10Base-2 network, each 185-meter cable length allows for only 30 computers. In a 10Base-5 network, there can be up to 100 computers for each 500-meter cable length. By adding several repeaters, you can extend the number of computers in either type of network.

Because of Ethernet bus limitations, a limit exists on how many LAN segments you can link. The 5-4-3 rule maximizes the number of computers on Ethernet bus networks. You can connect up to five LAN segments (using the maximum cable length) to four repeaters, allowing as many as three of the segments to have networked computers.

Therefore, a 10Base-2 network using the 5-4-3 rule can have 90 network nodes. A 10Base-5 network can have up to 300 network nodes. In contrast, the 10Base-T network that has to use network hubs to link network nodes together can have up to 1,024 network stations.

 Isolating LAN Segments. As Ethernet networks grow, the speed at which data moves through the network can slow down. To minimize this problem, network engineers use a network bridge. The network bridge only allows data destined for network nodes on its segment to cross the bridge from outside the segment. It also prevents data intended for network nodes on its segment from straying across the bridge in the rest of the network.

This makes for much better network performance. Network bridges operate on the data link layer of the OSI reference model and are able to pass non-routable network protocols such as NetBEUI.

The network switch can also reduce network bottlenecks. It does so by quickly linking two network nodes and then quickly linking another two nodes when the first connection is done. Network switches allow the creation of VPN(Virtual Private Networks) between network nodes. VPNs are basically private lines of communication that travel through the public Internet. They’re created by special software and hardware on the sending and receiving ends of the message that encrypts and decrypts network information as it enters and exits the Internet. This is a much more secure method of communication and makes things difficult for the wily hacker.  

by Bill Hayes

View the graphics that accompany this article.
Network Connections
How Client-Server Networks Work
(NOTE: These pages are PDF (Portable Document Format) files. You will need Adobe Acrobat Reader to view these pages. Download Adobe Acrobat Reader)

Want more information about a topic you found of interest while reading this article? Type a word or phrase that identifies the topic and click "Search" to find relevant articles from within our editorial database.

© Copyright by Sandhills Publishing Company 2001. All rights reserved.