Ethernet Communication Architecture

In the preceding section, you learned that every network has a physical and logical topology. The topology can provide important information on the capabilities of the network, from its reliability to its method of communication. The bus and star topologies are two such topologies that are used with the Ethernet standard for communication on a network.

Ethernet is one of the oldest network protocols and it is the most popular. Xerox first developed it in the 1970s. And in the 1980s, Xerox, Intel, and Digital Equipment Corporation proposed formal Ethernet specifications. It has its own standard from the Institute of Electrical and Electronic Engineers (IEEE), the IEEE 802.3 standard. The Ethernet protocol is used at the Physical and Media Access Control (MAC) layers of the OSI model (see Chapter 8, 'A Communications Framework').

Ethernet works with a contention scheme (a way of accessing the network) called Carrier Sense Multiple Access with Collision Detection (CSMA/CD). CSMA/CD works by allowing any computer to transmit at any time, assuming the line is free. The steps in this process are as follows:

Exchange/Sequenced Packet Exchange (IPX/SPX) is a proprietary protocol that Novell developed for use with its NetWare software product. Because it is proprietary to Novell, it is not as commonly used as TCP/IP.

1. When a workstation wants to send a packet on the network, the workstation listens to see whether any other nodes are transmitting packets over the network.

2. If the network is in use, the workstation waits.

3. If the network is not in use, the workstation sends its packet.

4. If two or more workstations send packets at the same time because they both thought the line was free, a collision occurs.

5. If a collision occurs, all workstations on the network cease transmission and each implements a back-off timer that generates a random number indicating how long the workstation must wait before it retransmits.

6. After the time expires, workstations can begin transmitting again.

   

Benefits

The benefits of Ethernet include the following:

• The protocol is fairly simple and does not have the overhead associated with Token Ring.

• As long as the network segment is not too busy, stations can transmit without any wait.

• Because Ethernet hardware is nonproprietary and less complicated than Token Ring hardware, Ethernet is less expensive.

Drawbacks

The drawbacks of Ethernet include the following:

• As network traffic increases, so does the probability of network collisions, which degrade network performance.

unshielded twisted-pair (UTP)

A type of media that can contain four, six, or eight wires. Pairs of wires are twisted together to prevent signal interference. The wires are then wrapped in a plastic cover. UTP is identified by the category nomenclature.

IEEE 802.3

The IEEE standard that is also known as Carrier Sense Multiple Access with Collision Detection (CSMA/CD) and defines how most Ethernet networks function.

IEEE 802.3u

The IEEE standard for 100Mbps Fast Ethernet. Defines the specifications for implementing Fast Ethernet at the Physical and Data-Link layers of the OSI model.

IEEE 802.3z

The IEEE standard for 1000Mbps Gigabit Ethernet over fiber-optic cable.

• Because all stations have equal access, there is no way to establish higher priority for nodes such as servers.

Ethernet Rules

It is estimated that about 70 percent of all networks use Ethernet. Ethernet is so popular because it is fairly easy to install and uses inexpensive hardware compared to Token Ring and FDDI. Although the latter two are well proven and reliable technologies, they are expensive.

Ethernet has the added advantage of offering higher-speed networks on the same wire. Two newer Ethernet technologies, Fast Ethernet and Gigabit Ethernet, perform at 100Mbps and 1000Mbps, respectively. That is a huge performance increase. And considering that most new computers come with network interface cards that support Fast Ethernet, it should be no surprise that Fast Ethernet is the default choice on new network installations.

The most popular form of Ethernet is 10BASE-T. With 10BASE-T Ethernet, you connect to the network with unshielded twisted-pair (UTP) cable; a connector on the end of the cable looks similar to, but is larger than, a telephone connector. One end of the cable attaches to the network card and the other end connects to a central hub called a concentrator.

Ethernet Standards

Ethernet has three primary standards:

IEEE 802.3  10Mbps Ethernet

IEEE 802.3u  100Mbps Fast Ethernet

IEEE 802.3z and IEEE 802.3ab  1000Mbps Gigabit Ethernet

Traditionally, people think of Ethernet as a 10Mbps standard. However, with the emerging standards that require networks to support voice, video, and other high-capacity bandwidth items, Fast Ethernet is now the standard for desktop computers and servers. Gigabit Ethernet is playing an increasing role as a key technology for supporting network backbones.

Ethernet has many other characteristics in addition to speed. Depending on which Ethernet standard you choose, you will have flexibility in which physical topology and which cable type you use.

Ethernet can use the following cable types:

• Thin coaxial cable (RG-58AU)

• Thick coaxial cable (RG-8 or RG-11)

• Unshielded twisted-pair (UTP) cables (type depends on access speed needed)

• Fiber-optic cable

Ethernet Naming Conventions

IEEE 802.3ab

The IEEE standard for 1000Mbps Gigabit Ethernet over unshielded twisted-pair cable (UTP).

coaxial

A type of media that has a single copper wire surrounded by plastic insulation, is wrapped in metal braid or foil, and is protected by a plastic cover.

fiber-optic

A type of media that uses glass or plastic to transmit light signals. Single-mode fiber-optic cable contains a single fiber. Multimode fiber-optic cable has two individually protected fibers.

American National Standards Institute (ANSI)

An organization that seeks to develop standardization within the computing industry. ANSI is the American representative to the ISO, or International Standards Organization.

The IEEE has defined naming conventions for Ethernet standards. The standards are to be named as follows:

• The first part of the name specifies the speed in megabits per second.

• The second part of the name specifies that the standard is using baseband (BASE) or broadband (BROAD) signaling.

• The last part of the name describes the type of cable being used, or an estimate of a maximum cable run for the standard.

For example, the 10BASE-T standard defines 10Mbps speed, using baseband signaling, over twisted-pair cabling.

This table summarizes the Ethernet standards.

Ethernet Standard	Speed in Mbps	Physical Topology	Cable Used
10BASE2	10	Bus	Thin coaxial cable RG-58AU (50-ohm cable)
10BASE5	10	Bus	Thick coaxial cable RG-8 or RG-11 (50-ohm cable)
10BASE-T	10	Star	Unshielded twisted-pair cable (category 3 or better)
10BASE-F	10	Star	Fiber-optic
100BASE-T	100	Star	Unshielded twisted-pair cable (Uses category 5 with all four pairs.)
100BASE-TX	100	Star	Unshielded twisted-pair cable
100BASE-FX	100	Star	Fiber-optic cable (Uses two strands of fiber as specified by ANSI.)
1000BASE-X	1000	Star	Fiber-optic and copper cable
1000BASE-T	1000	Star	Unshielded twisted-pair (UTP)

Ethernet Hardware

Depending on which Ethernet standard you choose, you can configure Ethernet as a physical bus or a physical star topology. This section provides an example of each topology and the hardware needed to support each configuration.

Ethernet Bus Topology

In the 1980s, before 10BASE-T became standardized, the most popular way of configuring Ethernet was in a physical bus topology. You might still use the bus topology in small networks, because it does not require a concentrator as the star configuration does. However, because the cost of Ethernet hardware has dropped drastically over the last 12 years, this is often not a large concern.

The most popular configuration for the bus topology uses the 10BASE2 standard, or thin coaxial cable. In this configuration, you connect a T-connector to the network card. You attach the cables to either side of the T-connector to form a linear bus network. At the ends of the bus, you need 50-ohm terminators. You should ground one end.

Ethernet Star Topology

The Ethernet star topology is by far the most commonly implemented topology for Ethernet networks. This topology is easier to set up and configure than the bus topology. It also provides more fault tolerance because a cable break does not cause the entire segment to go down.

In the star topology, you attach all your devices to the hub, or concentrator. You can buy concentrators with whatever number of ports you require, from 4 to 48 in a single unit.

When determining the hardware that you will use, you should first decide whether you will need 10Mbps, 100Mbps, or 1000Mbps networking capabilities. This factor determines the type of hub, Ethernet card, and cabling you will need. After your hardware is in place, the star will look like the following diagram.

In this graphic, both the workstations and the file server are connected to the network by using a 10BASE-T hub. In a small network of 5 to10 computers, this configuration is fine. But as the number of computers on the network grows, you will need to reconsider the design. Most network designers will select a hub that has at least a 100BASE-TX connection. The file server will connect to the 100BASE-TX to give optimum performance for multiple computers. More recently, networks are being designed with 100BASE-TX hubs to provide optimal network performance.