Why star topology is commonly preferred

LAN Topologies

Read Case Projects 3-1, 3-2, & 3-3 of the Guide to Networking Essentials.

Determine what type of physical and logical topology will be used for each scenario. Substantively explain the reasoning for your answer.

Create a visual representation of each chosen topology by copying and pasting the shapes provided in Appendix D—such as a bus, ring, star, or mesh—into a Microsoft® Word document. Note that you will need to use the Draw feature in Microsoft® Word to create lines between the shapes, and you may have to use some shapes more than once. Refer to the Figures 3-1 through 3-4 of Guide to Networking Essentials for examples.

chapter3

Network Topologies and Technologies

After reading this chapter and completing the exercises, you will be able to:

● Describe the primary physical networking topologies in common use ● Describe the primary logical networking topologies in common use ● Describe major LAN networking technologies

109 Copyright 2011 Cengage Learning. All Rights Reserved. May not be copied, scanned, or duplicated, in whole or in part. Due to electronic rights, some third party content may be suppressed from the eBook and/or eChapter(s).

Editorial review has deemed that any suppressed content does not materially affect the overall learning experience. Cengage Learning reserves the right to remove additional content at any time if subsequent rights restrictions require it.

Not so long ago, there was a real choice to be made between available network topologies and technologies when designing and building a new internetwork. Thankfully, this area of networking has gotten simpler rather than more complex, mainly because the choices have narrowed, with inferior or costly solutions becoming obsolete.

This chapter discusses network topologies, which describe both the physical arrangement of cabling or pathways between network devices and the logical manner in which data is trans- ferred from device to device. Next, you learn about network technologies or architectures that describe the methods computers use to transmit data to the networking medium in an orderly fashion. As you’ll see, the topology and technology are often tightly coupled, as certain technologies can be used only with certain topologies. The choices have been limited because only a few technologies and topologies remain as viable options. As is often the case, however, it helps to know where networking started to get an idea of where it might be heading. So even though some information covered in this chapter is obsolete or nearly so, your understanding of these older technologies will help you better understand current and future technologies.

Physical Topologies The word “topology,” for most people, describes the lay of the land. A topographic map, for example, shows the hills and valleys in a region, whereas a street map shows only the roads. A network topology describes how a network is physically laid out and how signals travel from one device to another. However, because the physical layout of devices and cables doesn’t necessarily describe how signals travel from one device to another, network topologies are categorized as physical and logical.

The arrangement of cabling and how cables connect one device to another in a network are considered the network’s physical topology, and the path data travels between computers on a network is considered the network’s logical topology. You can look at the physical topology as a topographic map that shows just the lay of the land along with towns, with only simple lines showing which towns have pathways to one another. The logical topology can be seen as a street map that shows how people actually have to travel from one place to another. As you’ll see, a network can be wired with one physical topology but pass data from machine to machine by using a different logical topology.

All network designs today are based on these basic physical topologies: bus, star, ring, and point-to-point. A bus consists of a series of computers connected along a single cable segment. Computers connected via a central device, such as a hub or switch, are arranged in a star topology. Devices connected to form a loop create a ring. Two devices connected directly to one another make a point-to-point topology. Keep in mind that these topologies describe the physical arrangement of cables. How the data travels along these cables might represent a dif- ferent logical topology. The dominant logical topologies in LANs include switching, bus, and ring, all of which are usually implemented as a physical star (discussed later in “Logical Topologies”).

Physical Bus Topology The physical bus topology, shown in Figure 3-1, is by far the simplest and at one time was the most common method for connecting computers. It’s a continuous length of cable

110 Chapter 3

Copyright 2011 Cengage Learning. All Rights Reserved. May not be copied, scanned, or duplicated, in whole or in part. Due to electronic rights, some third party content may be suppressed from the eBook and/or eChapter(s).

Editorial review has deemed that any suppressed content does not materially affect the overall learning experience. Cengage Learning reserves the right to remove additional content at any time if subsequent rights restrictions require it.

connecting one computer to another in daisy-chain fashion. One of this topology’s strengths is that you can add a new computer to the network simply by stringing a new length of cable from the last computer in the bus to the new machine. However, this strength is countered by a number of weaknesses:

● There’s a limit of 30 computers per cable segment. ● The maximum total length of cabling is 185 meters. ● Both ends of the bus must be terminated. ● Any break in the bus brings down the entire network. ● Adding or removing a machine brings down the entire network temporarily. ● Technologies using this topology are limited to 10 Mbps half-duplex communication

because they use coaxial cabling, discussed in Chapter 4.

Because of the preceding limitations, a physical bus topology is no longer a practical choice, and technology has moved past this obsolete method of connecting computers. However, the original Ethernet technology was based on this topology, and the basis of current LAN technol- ogy has its roots in the physical bus. So your understanding of bus communication aids your general understanding of how computers communicate with each other across a network.

How Data Travels in a Physical Bus Two properties inherent in a physical bus are signal propagation and signal bounce. In any network topology, computers communicate with each other by sending information across the media as a series of signals. When copper wire is the medium, as in a typical physical bus, these signals are sent as a series of electrical pulses that travel along the cable’s length in all directions. The signals continue traveling along the cable and through any connecting devices until they weaken enough that they can’t be detected or until they encounter a device that absorbs them. This traveling across the medium is called signal propagation. However, even if a signal encounters the end of a cable, it bounces back and travels in the other direction until it weakens or is otherwise impeded.

When a signal hits the end of a cable and bounces back up the cable’s length, it interferes with signals following it, much like an echo. Imagine if you were trying to communicate

Figure 3-1 A physical bus topology network Courtesy of Course Technology/Cengage Learning

3

Physical Topologies 111

Copyright 2011 Cengage Learning. All Rights Reserved. May not be copied, scanned, or duplicated, in whole or in part. Due to electronic rights, some third party content may be suppressed from the eBook and/or eChapter(s).

Editorial review has deemed that any suppressed content does not materially affect the overall learning experience. Cengage Learning reserves the right to remove additional content at any time if subsequent rights restrictions require it.

in an empty room with hard walls that caused your voice to echo continuously. The echo from the first words out of your mouth would garble the sound of words that followed, and your message would be unintelligible. The term used when electricity bounces off the end of a cable and back in the other direction is called signal bounce or reflection. To keep signal bounce from occurring, you do what you would to keep excessive echo from occurring; you install some type of material at both ends of the medium to absorb the signal. In a physical bus, you install a terminator, which is an electrical component called a resistor that absorbs the signal instead of allowing it to bounce back up the wire.

Physical Bus Limitations Now that you know more about how a physical bus works, the previous list of weaknesses needs some additional explanation. The limitation of 30 sta- tions per cable segment means only 30 computers can be daisy-chained together before the signal becomes too weak to be passed along to another computer. As an electrical signal encounters each connected workstation, some of its strength is absorbed by both the cabling and the connectors until the signal is finally too weak for a computer’s NIC to interpret. For the same reason, the total length of cabling is limited to 185 meters, whether there’s 1 con- nected station or 30 connected stations. The network can be extended in cable length and number of workstations by adding a repeater to the network, which, as you know, regener- ates the signal before sending it out.

At all times, both ends of the bus must be terminated. An unterminated bus results in signal bounce and data corruption. When a computer is added or removed from the network, both ends are no longer terminated, resulting in an interruption to network communication.

For a small network of only a few computers, you might think a bus topology is fine, until you consider the last weakness listed: maximum bandwidth of 10 Mbps half-duplex com- munication. A physical bus uses coaxial cable (a cabling type discussed in Chapter 4, similar to what’s used in cable TV connections), which is limited to a top speed of 10 Mbps and communication in only half-duplex mode. Most of today’s networks use twisted-pair cabling, which can operate at 100 Mbps or faster and run in full-duplex mode, so communi- cation between devices is much faster.

For all these reasons, the physical bus topology has long since fallen out of favor and been replaced largely by the star topology, discussed next.

Physical Star Topology The physical star topology uses a central device, such as a hub or switch, to interconnect computers in a LAN (see Figure 3-2). Each computer has a single length of cable going from its NIC to the central device.

Some advantages of a physical star topology are the following:

● Much faster technologies are used than in a bus topology. ● Centralized monitoring and management of network traffic is possible. ● Network upgrades are easier.

A physical star is the topology of choice for these reasons and more. With a central device, communication options are available that simply aren’t possible with a physical bus. For example, the central device can be a 100 Mbps hub, which increases a physical bus’s top

112 Chapter 3

Copyright 2011 Cengage Learning. All Rights Reserved. May not be copied, scanned, or duplicated, in whole or in part. Due to electronic rights, some third party content may be suppressed from the eBook and/or eChapter(s).

Editorial review has deemed that any suppressed content does not materially affect the overall learning experience. Cengage Learning reserves the right to remove additional content at any time if subsequent rights restrictions require it.

speed tenfold, or a switch, making it possible for multiple communication sessions to occur simultaneously and in full-duplex mode.

As a budding network administrator, being able to monitor and manage your network with a central device is a big advantage over what was possible with a physical bus topology. Today’s hubs and switches can include software that collects statistics about your network traffic patterns and even alerts you when excessive errors or unusually high traffic rates are occurring on your network. You don’t get these features in a $19.99 hub or switch, but enterprise-level devices can be equipped with several network management tools.

As long as your current cabling and installed NICs support it, your network can be upgraded quickly and easily from a ponderous 10 Mbps hub-based LAN to a blazing fast 100 Mbps or even 1000 Mbps switched network simply by replacing the central device. In addition, if your NICs must also be upgraded, you can upgrade in steps because most devices support multiple speeds. So if you want to upgrade from 100 Mbps to 1000 Mbps, you can replace the central device with a switch that supports both speeds, and then upgrade NICs as time and money allow. The switch transmits and receives on each port at the speed supported by the NIC connected to that port.

What happens if the number of workstations you need to connect exceed the number of ports on the central device? In this case, you can connect hubs or switches, as you learned in Chapter 2. When several hubs or switches must be connected, usually one device is used as the central connecting point, forming an extended star.

Extended Star The extended star topology, shown in Figure 3-3, is the most widely used in networks containing more than just a few computers. As the name implies, this topology is a star of stars. A central device, usually a switch, sits in the middle. Instead of attached computers forming the star’s arms, other switches (or hubs) are connected to the central switch’s ports. Computers and peripherals are then attached to these switches or

Switch

Figure 3-2 A physical star topology network Courtesy of Course Technology/Cengage Learning

3

Physical Topologies 113

Copyright 2011 Cengage Learning. All Rights Reserved. May not be copied, scanned, or duplicated, in whole or in part. Due to electronic rights, some third party content may be suppressed from the eBook and/or eChapter(s).

Editorial review has deemed that any suppressed content does not materially affect the overall learning experience. Cengage Learning reserves the right to remove additional content at any time if subsequent rights restrictions require it.

hubs, forming additional stars. The extended star is sometimes referred to as a “hierarchical star” because there are two or more layers of stars, all connecting back to the central star.

The extended star can be used to connect many computers, with the central device running at a very fast speed to shuttle data between the LAN’s outer stars. This topology is most effective when the center of the star is running at a much faster speed than other devices; for example, the central device can run at 1000 Mbps while other devices run at 100 Mbps.

How Data Travels in a Physical Star The details of how data travels from com- puter to computer in a physical star depend on the type of central device. Data transmission starts at a device at the end of one of the central device’s arms. From there, it travels along the network medium’s length until it arrives at the central device. As you know from learn- ing how hubs and switches work, the transmission path differs, depending on the device. Other devices, such as multistation access units (MAUs) used in token ring networks, move data differently. The type of central device, therefore, determines the logical topology, discussed later in this chapter.

Physical Star Disadvantages With all the clear advantages of a physical star, you might wonder whether there are any disadvantages. None outweigh the advantages, but it’s worth mentioning that the central device represents a single point of failure. In other words, if the hub or switch fails or someone kicks the power cord out of the outlet, down goes the entire network. Thankfully, these devices tend to be reliable and are usually placed out of the way of everyday foot traffic. That being said, they do fail from time to time, and having a spare on hand is a good idea.

When a physical bus was still the norm and the physical star was just coming on the net- working scene in the late 1980s, it was often argued that because each computer must be

Switch

Switch

SwitchSwitch

Switch

Figure 3-3 An extended star topology network Courtesy of Course Technology/Cengage Learning

114 Chapter 3

Copyright 2011 Cengage Learning. All Rights Reserved. May not be copied, scanned, or duplicated, in whole or in part. Due to electronic rights, some third party content may be suppressed from the eBook and/or eChapter(s).

Editorial review has deemed that any suppressed content does not materially affect the overall learning experience. Cengage Learning reserves the right to remove additional content at any time if subsequent rights restrictions require it.

cabled directly to the central device, instead of a bus’s daisy-chain arrangement, more cable was required to connect computers. This point is indeed true, and at the time, the amount of cabling needed was a factor in designing a network with a bus or star arrangement. By the time the star network’s advantages were fully realized in the mid-1990s, however, the cabling cost difference had diminished substantially, and the advantages clearly outweighed the minor cost disadvantage.

Physical Ring Topology A physical ring topology is like a bus, in that devices are daisy-chained one to another, but instead of terminating each end, the cabling is brought around from the last device back to the first device to form a ring. This topology had little to no following in LANs as a way to connect computers. It was used, however, to connect LANs with a technology called Fiber Distributed Data Interface (FDDI). FDDI was most often used as a reliable and fast network backbone, which is cabling used to communicate between LANs or between hubs or switches. In Figure 3-4, the devices used to connect buildings form a ring, but computers on each LAN are connected with a physical star topology.

The physical ring also had reliability issues because data had to be forwarded from one station to the next. Unlike a bus, inwhich data travels in all directions and is terminated at both ends, a ring doesn’t have any beginning or end. So each station must reproduce data and pass it along to the next station until it reaches the destination or the originator of the data. In other words, data always travels in one direction. If any station in the ring fails, data can no longer be passed along, and the ring is broken.

Technologies such as FDDI overcome some problems with a physical ring network by creating a dual ring, in which data can travel in both directions so that a single device failure doesn’t break the entire ring. However, this technology is costly, and although it was used extensively in the 1990s and early 2000s because it was fast (100 Mbps) and reliable, 100 Mbps and 1000 Mbps Ethernet have largely supplanted it with an extended star technology.

Building A Building B

Building C

LAN switch FDDI hub

Figure 3-4 A physical ring topology is usually used to connect LANs Courtesy of Course Technology/Cengage Learning

3

Physical Topologies 115

Copyright 2011 Cengage Learning. All Rights Reserved. May not be copied, scanned, or duplicated, in whole or in part. Due to electronic rights, some third party content may be suppressed from the eBook and/or eChapter(s).

Editorial review has deemed that any suppressed content does not materially affect the overall learning experience. Cengage Learning reserves the right to remove additional content at any time if subsequent rights restrictions require it.

Point-to-Point Topology As its name implies, a point-to-point topology is a direct link between two devices. It’s most often used in WANs, in which a device on a business’s network has a dedicated link to a telecommunication provider, such as the local phone company. The connection then hooks into the phone company’s network to provide Internet access or a WAN or MAN link to a branch office. The advantage of this type of topology is that data travels on a dedicated link, and its bandwidth isn’t shared with other networks. The disadvantage is that this topology tends to be quite expensive, particularly when used as a WAN link to a distant branch office.

Point-to-point topologies are also used with wireless networks in what’s called a wireless bridge. This setup can be used to connect two buildings without using a wired network (see Figure 3-5) or to extend an existing wireless network.

A rudimentary LAN can also be set up with a point-to-point topology by connecting a cable between the NICs on two computers. Of course, this method allows only two computers on the network, but it can be used effectively for transferring files from one computer to another in the absence of a hub or switch.

So as you can see, point-to-point topologies are used for specialized purposes. They aren’t commonly used in LANs; they’re used more often in WANs and large internetworks.

Mesh Topology Amesh topology connects each device to every other device in a network. You can look at a mesh topology as multiple point-to-point connections for the purposes of redundancy and fault tolerance. Figure 3-6 shows a full mesh topology between four locations, with the switch in each location providing connectivity to multiple computers. Each switch is connected to every other switch, which is called a “full mesh.” If each switch were connected to only two other switches, it would be called a “partial mesh.” In either case, the purpose of creating a mesh topology is to ensure that if one or more connections fail, there’s another path for reaching all devices on the network. For example, in Figure 3-6, two connections could fail, but all devices could still communicate with one another. This type of topology is used mostly commonly in large internetworks andWANs, where routers or switches in multiple buildings or towns are connected in a partial or full mesh. Parts of the Internet are also designed with a partial mesh topology, in which major ISPs are connected so that even if one ISP’s network fails, data can bypass this part of the network to get to its destination.

Mesh topologies, although reliable, are also expensive because of the additional cabling and ports required. In most cases, the ports used to connect devices are the highest speed available, such as 1 Gbps or 10 Gbps, and they often use expensive fiber-optic cabling for connecting buildings.

Figure 3-5 A point-to-point wireless topology Courtesy of Course Technology/Cengage Learning

116 Chapter 3

Copyright 2011 Cengage Learning. All Rights Reserved. May not be copied, scanned, or duplicated, in whole or in part. Due to electronic rights, some third party content may be suppressed from the eBook and/or eChapter(s).

Editorial review has deemed that any suppressed content does not materially affect the overall learning experience. Cengage Learning reserves the right to remove additional content at any time if subsequent rights restrictions require it.

Logical Topologies As mentioned, a network’s logical topology describes how data travels from computer to com- puter. In some cases, as with a physical bus and physical ring, the logical topology mimics the physical arrangement of cables. In other cases, as with a physical star, the electronics in the central device determine the logical topology.

A network’s logical topology reflects the underlying network technology (covered later in “Network Technologies”) used to transfer frames from one device to one another. Table 3-1 summarizes the main logical topologies, the technologies using them, and the physical topolo- gies for implementing them.

Chicago

Los Angeles Phoenix

New York

WAN link

Figure 3-6 Switches in each building are connected in a full mesh topology Courtesy of Course Technology/Cengage Learning

3

Table 3-1 Logical topologies and associated network technologies and physical topologies

Logical topology

Network technology

Physical topology Description

Bus Ethernet Bus or star A logical bus topology can be implemented as a physical bus (although this topology is nowobsolete).When a logical bus is implemented as a physical star usingwired Ethernet, the center of thestar is anEthernethub.Whatever thephysical topology is, data transmitted from a computer is received by all other computers.

Wireless LANs Star Wireless LANs use a physical star topology because they connect through a central access point. However, only one device can

Logical Topologies 117

Copyright 2011 Cengage Learning. All Rights Reserved. May not be copied, scanned, or duplicated, in whole or in part. Due to electronic rights, some third party content may be suppressed from the eBook and/or eChapter(s).

Editorial review has deemed that any suppressed content does not materially affect the overall learning experience. Cengage Learning reserves the right to remove additional content at any time if subsequent rights restrictions require it.

You have seen what a logical bus looks like when implemented as a physical bus. All computers are daisy-chained to one another, and network signals travel along the cable’s length in all directions, much like water flowing through interconnected pipes. When a logical bus is implemented as a physical star, the sameprocess occurs, but the pathways are hidden inside the central hub. Figure 3-7 shows what a logical bus might look like when implemented with a hub.

transmit at a time and all devices hear the transmission, so a wireless LAN can be considered a logical bus topology.

Ring Token ring Star Token ring networks use a central device called a multistation access unit (MAU or MSAU). Its electronics form a logical ring, so data is passed from computer to computer in order, until it reaches the destination device.

FDDI Ring As discussed, FDDI devices are connected in a physical ring, and data passes from device to device until it reaches the destination.

Switched Ethernet Star A switched logical topology using a physical star topology running Ethernet is by far themost common topology/technology combination now and likely will be well into the future. A switched topology creates dynamic connections or circuits between two devices whenever data is sent. This topology is sometimes considered a switched point-to-point topology because a circuit is established between two points as needed to transfer data (like turning on a switch), and then the circuit is broken when it’s no longer needed (like turning off a switch).

Signal Signal

Signal Signal

Logical bus inside a network hub

Figure 3-7 A logical bus implemented as a physical star Courtesy of Course Technology/Cengage Learning

Table 3-1 Logical topologies and associated network technologies and physical topologies (continued )

Logical

topology Network technology

Physical topology Description

118 Chapter 3

Copyright 2011 Cengage Learning. All Rights Reserved. May not be copied, scanned, or duplicated, in whole or in part. Due to electronic rights, some third party content may be suppressed from the eBook and/or eChapter(s).

Editorial review has deemed that any suppressed content does not materially affect the overall learning experience. Cengage Learning reserves the right to remove additional content at any time if subsequent rights restrictions require it.

A logical bus is sometimes called a “shared media topology” because all stations must share the bandwidth the media provides.

A logical ring using a physical star implements the ring inside the central device’s electronics, which is an MAU in the token ring technology. Data is passed from one node or computer to another until it reaches the destination device (see Figure 3-8). When a port has no device connected to it, it’s simply bypassed, and data is sent out the next connected port.

A switched topology works something like what’s shown in Figure 3-9. Although there’s always an electrical connection between the computer and switch, when no data is being transferred, there’s no logical connection or circuit between devices. However, when the switch receives a frame, a logical circuit is made between the source and destination devices until the frame is transferred.

Figure 3-8 A logical ring implemented as a physical star Courtesy of Course Technology/Cengage Learning

3

PC 6PC 5PC 4

No packets being transmitted

PC 1 PC 2

PC 1 and PC 6 communicate while PC 2 and PC 5 communicate

PC 3

Figure 3-9 The logical functioning of a switch Courtesy of Course Technology/Cengage Learning

Logical Topologies 119

Copyright 2011 Cengage Learning. All Rights Reserved. May not be copied, scanned, or duplicated, in whole or in part. Due to electronic rights, some third party content may be suppressed from the eBook and/or eChapter(s).

Editorial review has deemed that any suppressed content does not materially affect the overall learning experience. Cengage Learning reserves the right to remove additional content at any time if subsequent rights restrictions require it.

To better understand how these logical topologies work, it helps to know the network technology that drives each topology (discussed later in “Network Technologies”).

Hands-On Project 3-1: Building a Physical Star Topology Network

Time Required: 20 minutes

Objective: Build a physical star topology network.

Required Tools/Equipment: Three workstations named Computer1, Computer2, and Com- puter3; a hub; and three patch cables. Workstations should be configured with an IP address or automatic IP address assignment. Each station should have Wireshark installed.

Description: In this project, you build a small physical star topology; this task can be done in groups of three or more or as an instructor demonstration. After each station is connected to the hub, you ping another station to verify connectivity. Next, you use Wireshark to cap- ture ping packets so that you can determine the network’s logical topology.

1. Power on the hub.

2. Connect each workstation to the hub with the supplied cables.

3. Inspect the hub and the workstation NIC to verify that you have a good connection with the hub. Write down how you determined whether the connection with the hub is good:

4. On each workstation, open a command prompt window, and then type ipconfig and press Enter to determine your IP address. Write down the IP address of each computer: ● IP address of Computer1:

● IP address of Computer2:

● IP address of Computer3:

5. Ping each computer to verify that you can communicate with it. If the pings aren’t suc- cessful, check that the IP addresses you wrote down are correct and the connection with the hub is good, and then try again.

6. Make sure you coordinate the rest of the project, starting with this step, with students at the other computers. Start Wireshark, and start a capture session by clicking the inter- face name listed in the Interface List section.

7. At the command prompt, ping the next computer. For example, if you’re at Computer1, ping Computer2; if you’re at Computer2, ping Computer3; and if you’re at Computer3, ping Computer1.

120 Chapter 3

Copyright 2011 Cengage Learning. All Rights Reserved. May not be copied, scanned, or duplicated, in whole or in part. Due to electronic rights, some third party content may be suppressed from the eBook and/or eChapter(s).

Editorial review has deemed that any suppressed content does not materially affect the overall learning experience. Cengage Learning reserves the right to remove additional content at any time if subsequent rights restrictions require it.

8. Based on which packets Wireshark captured, what’s your logical topology?

9. Exit Wireshark, close all open windows, and leave the computers running if you’re con- tinuing to the next project.

Network Technologies A network technology, as the phrase is used here, can best be described as the method a net- work interface uses to access the medium and send data frames and the structure of these frames. Other terms include network interface layer technologies, network architectures, and Data Link layer technologies. What it comes down to is whether your network uses Ethernet, 802.11 wireless, token ring, or some combination of these and other technologies to move data from device to device in your network. Most LANs are now based on a combination of Ethernet and 802.11 wireless. WANs use technologies specifically designed to carry data over longer distances, such as frame relay, FDDI, Asynchronous Transfer Mode (ATM), and others.

The network technology sometimes, but not always, defines frame format and which media types can be used to transfer frames. For example, different Ethernet speeds specify a mini- mum grade of copper or fiber-optic cabling that must be used as well as the connectors attached to the ends of cables. FDDI requires fiber-optic cabling, but other technologies, such as frame relay, can run on a variety of media types.

This book focuses on LAN technologies with particular emphasis on Ethernet and 802.11 wireless because they’re the most commonly used. Some WAN technologies are also described briefly in this chapter and in more detail in Chapter 12.

Network Technologies and Media Because some of the network technologies discussed in this chapter specify the types of media they require to operate, the following sections summarize the most common media types. However, you can find more details on network media in Chapter 4.

Unshielded Twisted Pair Unshielded twisted pair (UTP) is the most common media type in LANs. It consists of four pairs of copper wire, with each pair tightly twisted together and contained in a plastic sheath or jacket (Figure 3-10).

3

Sheath

Figure 3-10 UTP cabling Courtesy of Course Technology/Cengage Learning

Network Technologies 121

Copyright 2011 Cengage Learning. All Rights Reserved. May not be copied, scanned, or duplicated, in whole or in part. Due to electronic rights, some third party content may be suppressed from the eBook and/or eChapter(s).

Editorial review has deemed that any suppressed content does not materially affect the overall learning experience. Cengage Learning reserves the right to remove additional content at any time if subsequent rights restrictions require it.

UTP comes in numbered categories, up to Category 7 as of this writing. The higher the category, the higher the cable’s bandwidth potential. Category 5 Enhanced (Cat 5E) and Category 6 (Cat 6) are the most common in wired LANs, allowing speeds up to 10 Gbps. UTP cabling is used in physical star networks, and the maximum cable length from NIC to hub or switch is 100 meters in LAN applications. UTP cabling is susceptible to electrical interference, which can cause data corruption, so it shouldn’t be used in electrically noisy environments.

Fiber-Optic Cabling Fiber-optic cabling uses extremely thin strands of glass to carry pulses of light long distances and at high data rates. It’s usually used in large internetworks to connect switches and routers and sometimes to connect high-speed servers to the network. Because of its capability to carry data over long distances (several hundred to several thousand meters), it’s also used in WAN applications frequently. Fiber-optic cabling isn’t susceptible to electrical interfer- ence, so unlike UTP, it can be used in electrically noisy environments. It requires two strands of fiber to make a network connection: one for transmitting and one for receiving.

Coaxial Cable Best known for its use in cable TV, coaxial cable is obsolete as a LAN medium, but it’s used as the network medium for Internet access via cable modem. Coaxial cable was the original media used by Ethernet in physical bus topologies, but its limitation of 10 Mbps half-duplex communication made it obsolete for LAN applications after star topologies and 100 Mbps Ethernet became the dominant standard. Coaxial cable in LANs can have lengths of around 200 meters.

Baseband and Broadband Signaling Network technologies can use media to transmit signals in two main ways: baseband and broadband. The baseband transmission method sends digital signals in which each bit of data is represented by a pulse of electricity (on copper media) or light (on fiber-optic media). These signals are sent at a single fixed frequency, using the medium’s entire bandwidth. In other words, when a frame is sent to the medium, it occupies the cable’s entire bandwidth, and no other frames can be sent along with it—much like having cable TV that carries only a single channel. LAN technolo- gies, such as Ethernet and token ring, use baseband transmission. If cable TV used baseband signaling, you would need one cable for each channel!

Thankfully, cable TV and cable modem Internet access use broadband transmission. Instead of digital pulses, broadband systems use analog techniques to encode binary 1s and 0s across a continuous range of values. Broadband signals move across the medium in the form of continuous electromagnetic or optical waves rather than discrete pulses. On broad- band systems, signals flow at a particular frequency, and each frequency represents a chan- nel of data. That’s why broadband systems, such as cable TV and Internet, can carry dozens or hundreds of TV channels plus Internet access on a single cable wire: Each channel oper- ates at a different frequency. In addition, incoming and outgoing Internet data use separate channels operating at different frequencies from TV channels.

Ethernet Networks Ethernet, the most popular LAN technology, has many advantages, including ease of installa- tion, scalability, media support, and low cost. It supports a broad range of transmission speeds, from 10 Mbps to 10 Gbps.

As discussed, Ethernet can operate in a bus or star physical topology and a bus or switched logical topology. It has been in use since the mid-1970s but didn’t mature as a technology until the early to mid-1980s. Ethernet being around for almost 40 years is a testament to the

122 Chapter 3

Copyright 2011 Cengage Learning. All Rights Reserved. May not be copied, scanned, or duplicated, in whole or in part. Due to electronic rights, some third party content may be suppressed from the eBook and/or eChapter(s).

Editorial review has deemed that any suppressed content does not materially affect the overall learning experience. Cengage Learning reserves the right to remove additional content at any time if subsequent rights restrictions require it.

original designers, whose forethought enabled Ethernet to scale from a 3 Mbps technology in its early years to a 10 Gbps and beyond technology today.

Although there are many variations of Ethernet, all forms are similar in their basic operation and frame formatting. What differs in the variations are the cabling, speed of transmission, and method by which bits are encoded on the medium. Because the frame formatting is the same, however, Ethernet variations are compatible with one another. That’s why you often see NICs and Ethernet hubs and switches described as 10/100 or 10/100/1000 devices. These devices can support multiple Ethernet speeds because the underlying technology remains the same, regardless of speed.

Ethernet Addressing Every Ethernet station must have a physical or MAC address. As you learned in Chapter 2, a MAC address is an integral part of network interface elec- tronics and consists of 48 bits expressed as 12 hexadecimal digits. When a frame is sent to the network medium, it must contain both source and destination MAC addresses. When a network interface detects a frame on the media, the NIC reads the frame’s destination address and compares it with its own MAC address. If they match or if the destination address is the broadcast MAC address (all binary 1s or FF:FF:FF:FF:FF:FF in hexadecimal), the NIC reads the frame and sends it to the network protocol for further processing.

Ethernet Frames A frame is the unit of network information NICs and switches work with. It’s the NIC’s responsibility to transmit and receive frames and a switch’s responsibil- ity to forward frames out the correct switch port to get the frame to its destination.

Ethernet frames come in four different formats, or frame types, depending on the network protocol used to send frames, and unfortunately, these frame types are incompatible with one another. They were developed during Ethernet’s early days, before standards were solid- ified. If your network needed to support multiple protocols, such as TCP/IP, IPX/SPX, and AppleTalk, you had to make sure your computers were configured to support all these frame types. Thankfully, TCP/IP has become the dominant network protocol in LANs, so supporting multiple frame types is largely unnecessary, except for networks that still run older Novell NetWare servers. Given this reality, this section examines only the frame type used by TCP/IP: Ethernet II. The other frame types are Ethernet SNAP, Ethernet 802.3, and Ethernet 802.2. For information on these frame types, see Appendix B.

The four Ethernet frame types are incompatible in the same Ether- net standard (such as using both Ethernet II and Ethernet SNAP in 100 Mbps), but each frame type is compatible with the same frame type in different standards. For example, Ethernet II in 10 Mbps Ethernet is compatible with Ethernet II in 100 Mbps and 1000 Mbps Ethernet.

Regardless of frame type, Ethernet networks can accommodate frames between 64 bytes and 1518 bytes. Shorter or longer frames are considered errors. Each frame is composed of the following (see Figure 3-11):

● A 14-byte frame header composed of these three fields:

● A 6-byte Destination MAC Address field ● A 6-byte Source MAC Address field ● A 2-byte Type field

3

Network Technologies 123

Copyright 2011 Cengage Learning. All Rights Reserved. May not be copied, scanned, or duplicated, in whole or in part. Due to electronic rights, some third party content may be suppressed from the eBook and/or eChapter(s).

Editorial review has deemed that any suppressed content does not materially affect the overall learning experience. Cengage Learning reserves the right to remove additional content at any time if subsequent rights restrictions require it.

● A Data field from 46 to 1500 bytes ● A frame trailer (frame check sequence [FCS]) of 4 bytes

You’ve already learned the purpose and format of destination and source MAC addresses. The Type field in the frame header indicates the network protocol in the data portion. For example, this field might indicate that the Data field contains an IP, IPv6, or ARP packet, to name just a few possibilities. The data portion, often referred to as the “frame payload,” contains network protocol header information as well as the actual data an application is transferring. The FCS in the frame trailer is an error-checking code (discussed later in “Ethernet Error Handling”).

There are exceptions to the 1518-byte maximum frame size. For exam- ple, a function of some switches requires an additional 4-byte field in the Ethernet frame, bringing the maximum size to 1522 bytes. In addi- tion, Jumbo frames of up to 9000 bytes are supported by some NICs and switches but aren’t officially supported in the current Ethernet stan- dards. To use Jumbo frames, the feature must be enabled on every device on the LAN and be implemented the same way by these devices.

Ethernet Media Access Before a NIC can transmit data to the network medium, it must adhere to some rules governing how and when the medium can be accessed for trans- mission. The rules ensure that data is transmitted and received in an orderly fashion and all stations have an opportunity to communicate. The set of rules for each networking technol- ogy is referred to as its media access method or media access control. Note that the acronym for “media access control” is MAC, which is where the term “MAC address” comes from.

The media access method Ethernet uses in half-duplex mode is Carrier Sense Multiple Access with Collision Detection (CSMA/CD). To understand this method better, break this term down into parts. “Carrier sense” means to listen. The rules for half-duplex Ethernet state that a device can send or receive data but can’t do both simultaneously. So before a device can send, it must listen to see whether the medium is already busy, much like a group of people having a conversation. Each person listens for a pause in the conversation before speaking up. “Multiple access” simply means that multiple computers can be listen- ing and waiting to transmit at the same time, which brings you to “collision detection.” A collision occurs if two or more devices on the same medium transmit simultaneously. For example, if two people are waiting to chime in on a group conversation, they both hear a lull in the conversation at the same time and speak up simultaneously, causing a “collision” in the conversation. Ethernet’s collision detection method is much like a person’s; Ethernet detects, or “hears,” the other station transmit, so it knows a collision has occurred. The NIC then waits for a random period before attempting to transmit again. Ethernet repeats the “listen before transmitting” process until it transmits the frame without a collision. Sim- ulation 7 on the book’s CD shows a simulation of the CSMA/CD process.

Destination MAC Address (6 bytes)

Source MAC Address (6 bytes) Type (2 bytes)

Data (46–1500 bytes) FCS (4 bytes)

Frame header Data (frame payload)

Frame trailer

Courtesy of Course Technology/Cengage Learning

Figure 3-11 Ethernet II frame format

124 Chapter 3

Copyright 2011 Cengage Learning. All Rights Reserved. May not be copied, scanned, or duplicated, in whole or in part. Due to electronic rights, some third party content may be suppressed from the eBook and/or eChapter(s).

Editorial review has deemed that any suppressed content does not materially affect the overall learning experience. Cengage Learning reserves the right to remove additional content at any time if subsequent rights restrictions require it.

Simulation 7: Ethernet operation using CSMA/CD

As you determined in Hand-On Project 2-4, when you attempted to create enough traffic to generate a collision, the CSMA/CD access method is efficient. It takes quite a bit of traffic to generate collisions, especially on a 100 Mbps network. However, the more devices on a logical bus topology and the more data they transmit, the greater the chance of a collision. So although CSMA/CD works well, today’s multimedia-heavy networks have somewhat outgrown it, and Ethernet has adapted to this development.

CSMA/CD is considered a contention-based access method, which means computers are allowed to send whenever they have data ready to send. Obviously, CSMA/CD modifies this rule somewhat by stipulating that the computer must listen first to ensure that no other station is in the process of transmitting.

Collisions and Collision Domains Remember that collisions can occur only in an Ethernet shared-media environment, which means a logical bus topology is in use. In this environment, all devices interconnected by one or more hubs hear all signals generated by all other devices. The signals are propagated from hub to hub until there are no more devices or until a device is encountered that doesn’t use a logical bus topology, such as a switch or a router. The extent to which signals in an Ethernet bus topology network are propagated is called a collision domain. Figure 3-12 shows a network diagram with two

3

Switch

Hub 2

Hub 1

Hub 3

Collision domain Collision domain

Hub 4 Hub 6

Hub 5

Figure 3-12 A network diagram showing two collision domains delimited by a switch Courtesy of Course Technology/Cengage Learning

Network Technologies 125

Copyright 2011 Cengage Learning. All Rights Reserved. May not be copied, scanned, or duplicated, in whole or in part. Due to electronic rights, some third party content may be suppressed from the eBook and/or eChapter(s).

Editorial review has deemed that any suppressed content does not materially affect the overall learning experience. Cengage Learning reserves the right to remove additional content at any time if subsequent rights restrictions require it.

collision domains enclosed in circles. All devices in a collision domain are subject to the possi- bility that whenever a device sends a frame, a collision might occur with another device send- ing a frame at the same time. This fact has serious implications for the number of computers that can reasonably be installed in a single collision domain. The more computers, the more likely it is that collisions occur. The more collisions, the slower network performance is.

Notice in Figure 3-12 that all computers connected to Hubs 1 to 3 are in the same collision domain, and computers connected to Hubs 4 to 6 are in a different collision domain. This is because a switch port delimits the collision domain, which means collisions occurring in one collision domain don’t propagate through the switch.

Although collisions in an Ethernet network are usually associated with hubs, technically it’s possible for a collision to occur with a computer connected to a switch. A collision with a switch can occur only if the NIC connected to the switch port is operating in half-duplex mode. In addition, the collision domain is limited to only the devices connected to a single switch port. The same is true of routers. However, given that an Ethernet frame of maxi- mum size is transmitted on a 10 Mbps switch in just over a millisecond and just over a microsecond on a 100 Mbps switch, the likelihood of a collision with a switch is low.

If a hub is connected to a switch port in an extended star topology, collisions can occur between devices connected to the hub and the switch port. To avoid collisions altogether, use only switches in your network design with computers that have NICs operating in full- duplex mode.

Ethernet Error Handling One reason for Ethernet’s low cost and scalability is its sim- plicity. It’s considered a best-effort delivery system, meaning that when a frame is sent, there’s no acknowledgement or verification that the frame arrived at its intended destina- tion. Ethernet relies on network protocols, such as TCP/IP, to ensure reliable delivery of data. It’s similar to the package delivery guy at a corporation. His job is to take what he’s given to its intended destination; it’s the package receiver’s job to verify its contents and let the sender know it was received.

Ethernet can also detect whether a frame has been damaged in transit. The error-checking code in an Ethernet frame’s trailer is called a Cyclic Redundancy Check (CRC), which is the result of a mathematical algorithm computed on the frame data. The CRC is calculated and placed in the frame trailer before the frame is transmitted. When the frame is received, the calculation is repeated. If the results of this calculation don’t match the CRC in the frame, it indicates that the data was altered in some way, usually from electrical interference. If a frame is detected as damaged, because Ethernet is a best-effort delivery system, it simply discards the frame but doesn’t inform the sending station that an error occurred. Again, it’s the network protocol’s job to ensure that all expected data was actually received. The net- work protocol or, in some cases, the application sending the data is responsible for resend- ing damaged or missing data, not Ethernet.

A collision is the exception to Ethernet’s lack of action when an error occurs. When frames are involved in a collision, Ethernet resends them automatically because all stations detect that a collision has occurred.

126 Chapter 3

Copyright 2011 Cengage Learning. All Rights Reserved. May not be copied, scanned, or duplicated, in whole or in part. Due to electronic rights, some third party content may be suppressed from the eBook and/or eChapter(s).

Editorial review has deemed that any suppressed content does not materially affect the overall learning experience. Cengage Learning reserves the right to remove additional content at any time if subsequent rights restrictions require it.

Half-Duplex Versus Full-Duplex Communication As discussed in Chapter 2, half-duplex communication means a station can transmit and receive data but not at the same time, much like a two-way radio. When Ethernet is implemented as a logical bus topology (using hubs), NICs can operate only in half-duplex mode and must use the CSMA/CD access method.

However, a network switch allows half-duplex or full-duplex communication. If a NIC is operating in half-duplex mode while connected to a switch, it must use CSMA/CD. How- ever, the only time a collision can occur in this circumstance is if the switch happens to transmit a frame to the NIC at the same time the NIC is attempting to transmit.

Full-duplex mode, by definition, means a NIC can transmit and receive simultaneously. Therefore, when an Ethernet NIC is operating in full-duplex mode connected to a switch, CSMA/CD isn’t used because a collision can’t occur in full-duplex mode. Because full- duplex mode eliminates the delays caused by CSMA/CD and allows double the network bandwidth, most Ethernet LANs now operate in this mode using switches.

Ethernet Standards Ethernet can operate at different speeds over different types of media, and each variation is associated with an IEEE standard. The following sections discuss many of these standards, some of which are obsolete or had limited use.

Standards Terminology Ethernet standards are generally expressed in one of two ways. One way is using the IEEE document number defining the standard. For example, IEEE 802.3 is the parent document specification for 10 Mbps Ethernet using thick coaxial cable, which was ratified in 1983. All other variations and speeds of Ethernet are subdocu- ments of the original 802.3 specification.

The second way of expressing an Ethernet standard is to use the XBaseY terminology. Most IEEE 802.3 documents describe the transmission speed, type of transmission, and length or type of cabling and are designated with terms such as 100BaseT. In 100BaseT, for example, the “100” designates the speed of transmission (100 Mbps), the “Base” indicates a base- band signaling method, and the “T” specifies twisted-pair cabling. All the BaseT Ethernet standards use a physical star topology. The following sections discuss the major standards and their designations.

10BaseT Ethernet 10BaseT Ethernet, defined by IEEE 802.3i, has been the mainstay of Ethernet networks since the early 1990s. It runs over Category 3 or higher UTP cabling and uses two of the four wire pairs. Because of its slower transmission speed, 10BaseT networks using a logical bus topology (with hubs) are more susceptible to collisions than faster 100BaseT networks are. In addition, the amount of data sent and received by a typical user makes 10BaseT seem slow in typical media-heavy environments compared with the more common 100BaseT and 1000BaseT standards.

If you work for an organization still using hubs, you need to know that there are limits to how many hubs you can string together to connect all computers. The rule for expanding a 10BaseT network with hubs is that no more than four hubs can be placed between two communicating workstations. This rule ensures that all stations on the network can detect a collision. Because of the limited time for signals to propagate through a network, if more

3

Network Technologies 127

Copyright 2011 Cengage Learning. All Rights Reserved. May not be copied, scanned, or duplicated, in whole or in part. Due to electronic rights, some third party content may be suppressed from the eBook and/or eChapter(s).

Editorial review has deemed that any suppressed content does not materially affect the overall learning experience. Cengage Learning reserves the right to remove additional content at any time if subsequent rights restrictions require it.

than four hubs exist between end stations, a collision on one end of the network might not be detected by stations on the other side of the network in time for them to react properly. If switches rather than hubs are used, there’s no such limitation because a collision on a switch can take place only between the switch and a single workstation.

A business network still using 10BaseT should upgrade to 100 or 1000BaseT to take full advan- tage of current technology. A home or small-office network that uses the network mainly for sharing Internet access and transferring documents can still use 10BaseT effectively if its Internet connection is considerably slower than 10 Mbps. However, 10BaseT is essentially an obsolete technology, and networks using 10BaseT should upgrade as soon as circumstances permit.

100BaseTX Ethernet 100BaseTX (often referred to as simply “100BaseT”), defined by IEEE 802.3u, is the most commonly used Ethernet variety today. It runs over Category 5 or higher UTP cable and uses two of the four wire pairs: one to transmit data and the other to receive data. There are other varieties of 100BaseT Ethernet (discussed later in this section), but 100BaseTX is the standard that’s usually in mind when discussing 100 Mbps Ethernet. It’s also sometimes called Fast Ethernet.