Most LANs use some form of cable as their network medium. Although there are many types of wireless media, cables are more reliable and generally provide faster transmission speeds than other media. Data-link layer protocols oftenprovide more than one cable specification to choose from. Each specificationincludes the type of cable to use, the cable grade, and the basic guidelines forinstalling it. The type of cable you choose should be based on the requirements of your installation, the nature of the site where your network is to be installed, and, of course, your budget.
As explained in Chapter 1, "Networking Basics," the topology of a network is the pattern used to connect the computers and other devices with the cable or other network medium. The Network+ exam always contains questions about the basic network topologies and their properties. The topology of your network is directly related to the type of cable you use. You cannot select a particular type of cable and install it using just any topology. However, you can create individual LANs using a different cable and topology for each LAN and connect them together using devices such as bridges, switches, and routers. When choosing the components with which to build a LAN, the topology should be one of the most important criteria you use to select a cable type. The three primary topologies used to build LANs are as follows:
You should also be familiar with the following additional topologies:
A network that uses the bus topology is one in which the computers are connected in a single line, with each system cabled to the next system. Bus networks are illustrated in Figure 2.1. Early Ethernet systems used the bus topology with coaxial cable, a type of network that is rarely seen today. The cabling of a bus network can take two forms: thick and thin. Thick Ethernet networks use a single length of coaxial cable with computers connected to it using smaller individual cables called Attachment Unit Interface (AUI) cables (sometimes called transceiver cables), as shown on the top half of Figure 2.1. Thin Ethernet networks use separate lengths of a narrower type of coaxial cable, and each length of cable connects one computer to the next, as shown in the bottom half of Figure 2.1.
The transceiver is an integral component of the network interface responsible for both transmitting and receiving data over the network medium. Thick Ethernet is the only form of Ethernet network that uses a transceiver that's separate from the network interface adapter. The transceiver itself connects to thecoaxial cable using a device called a vampire tap, named for the metal teeth with which it penetrates the cable sheath to make a connection with the copper conductor inside. The transceiver is then connected to the network interface adapter in the computer using an AUI (transceiver) cable. All of the other Ethernet physical layer standards have their transceivers integrated into the network interface adapter card and do not require separate AUI cables.
Figure 2.1 Bus topology cabling options
When any one of the computers on the network transmits data, the signals travel down the cable in both directions, reaching all of the other systems. A bus network always has two open ends, which must be terminated. Termination is the process of installing a resistor pack at each end of the bus to negate the signals that arrive there. Without terminators, the signals reaching the end of the bus would reflect back in the other direction and interfere with the newer signals being transmitted.
Run the BusTopology video located in the Demos folder on the CD-ROMaccompanying this book for a demonstration of bus topology communications, signal bounce, and termination.
The main problem with the bus topology is that a single faulty connector, faulty terminator, or break in the cable affects the functionality of the entire network. Signals that cannot pass beyond a certain point on the cable cannot reach all of the computers beyond that point. In addition, when a component failure splits the network into two segments, each half of the cable is also unterminated. On the half of the network that does receive the signals transmitted by each computer, signal reflection garbles the data. This is one of the primary reasons that bus networks are rarely used now.
Run the BusFailure video located in the Demos folder on the CD-ROM accompanying this book for a demonstration of a bus topology failure.
Whereas the bus topology has the computers in a network connected directly to each other, the star topology uses a central cabling nexus called a hub or concentrator. In a star network, each computer is connected to the hub using a separate cable, as shown in Figure 2.2. Most of the Ethernet LANs installed today, and many LANs using other protocols as well, use the star topology. Star LANs can use several different cable types, including various types of twisted-pair and fiber optic cable.
Figure 2.2 The star topology uses an individual connection for each computer to provide a greater measure of fault tolerance than the bus topology
The unshielded twisted pair (UTP) cables used on most Ethernet LANs are usually installed using a star topology. Functionally, a star network uses a shared network medium, just as a bus network does. Despite the fact that each computer connects to the hub with its own cable, the hub propagates all signals entering through its ports out through all of its other ports. Signals transmitted by one computer are therefore received by all other computers on the LAN.
The main advantage of the star topology is that each computer has its own dedicated connection to the hub, providing the network with a measure of fault tolerance. If a single cable or connector should fail, only the computer connected to the hub by that cable is affected. The disadvantage of the star topology is that an additional piece of hardware, the hub, is required to implement it. If the hub should fail, the entire network goes down. However, this is a relatively rareoccurrence because hubs are relatively simple devices that are usually found in a protected environment, such as a data center or server closet.
Run the StarTopology video located in the Demos folder on the CD-ROMaccompanying this book for a demonstration of the star topology.
It might seem as though the size of an Ethernet network using the star topology is limited to the number of ports in the hub. However, if your network growsuntil all of the hub ports are filled, you can still expand it by adding a second hub, and in some cases, a third and a fourth. To add a second hub to a star network, connect it to the first using a standard cable and a special port in one ofthe hubs called an uplink port. This creates what is known as a hierarchical star topology (sometimes known as a branching tree network), shown in Figure 2.3. A standard 10-Mbps Ethernet network can support up to four hubs connectedin this fashion, but a Fast Ethernet network can generally support only two.
Figure 2.3 A hierarchical star network uses two or more interconnected hubs
In terms of signal transmissions, a ring network is like a bus in that each computer is logically connected to the next. However, in a ring network, the two ends are connected instead of being terminated, thus forming an endless loop. This enables a signal originating on one computer to travel around the ring to all of the other computers and eventually back to its point of origin. Networks such as Token Ring, which use token passing for their Media Access Control (MAC) mechanism (as explained in Lesson 2: The OSI Reference Model, in Chapter 1, "Networking Basics"), are wired using a ring topology. The most important thing to understand about the ring topology is that, in most cases, it is strictly a logical construction, not a physical one. To be more precise, the ring exists in the wiring of the network, but not in the cabling.
A cable is a device that contains a number of signal conductors, usually in the form of separate wires. A twisted-pair cable, for example, contains eight individual wires within a single sheath.
When you look at a network that uses the ring topology, you may be puzzled to see what looks like a star. In fact, the cables for a ring network connect to a hub and take the form of a star. The ring topology is actually implemented logically, using the wiring inside the cables (see Figure 2.4). Ring networks use a special type of hub, called a multistation access unit (MAU), which receives data through one port and transmits it out through each of the others in turn (not simultaneously, as with an Ethernet hub). For example, when the computer connected to port number 3 in an eight-port MAU transmits a data packet, the MAU receives the packet and transmits it out through port number 4 only. When the computer connected to port number 4 receives the packet, it immediately returns it to the MAU, which then transmits it out through port number 5, and so on. This process continues until the MAU has transmitted the packet to each computer on the ring. Finally, the computer that generated the packet receives it back again and is then responsible for removing it from the ring. If you were to remove the wire pairs from the sheaths of the cables that make up a ring network, you would have a circuit that runs from the MAU to each computer and back to the MAU.
Figure 2.4 A ring network uses a ring topology in a logical sense only; the cables are actually arranged in the form of a star
Run the RingTopology video located in the Demos folder on the CD-ROMaccompanying this book for a demonstration of the ring topology.
The design of the physical star topology used by the ring makes it possible for the network to function even when a cable or connector fails. The MAU contains special circuitry that removes a malfunctioning workstation from the ring, but still preserves the logical topology. By comparison, a network that is literally cabled as a ring would have no MAU, but a cable break or connector failure would cause the network to stop functioning completely. The one commonly used protocol that does include an option for a physical ring topology, Fiber Distributed Data Interface (FDDI), defines the use of a double ring, which consists of two separate physical rings with traffic flowing in opposite directions. When computers are connected to both rings, the network can still function despite a cable failure.
Run the RingFailure video located in the Demos folder on the CD-ROMaccompanying this book for a demonstration of a ring topology failure.
The mesh topology, in the context of local area networking, is more of a theoretical concept than an actual real-world solution. On a mesh LAN, each computer has a dedicated connection to every other computer, as shown in Figure 2.5. In reality, this topology only exists on a two-node network. For a mesh network with three computers or more, it would be necessary to equip each computer with a separate network interface for every other computer on the network. Thus, for a five-node network, each computer would require four network interface adapters, which is certainly not practical. A mesh LAN provides excellent fault tolerance, however, as there is no single point of failure that can affect more than one computer.
Figure 2.5 A mesh LAN uses a separate cable connection for each pair of computers
In internetworking, the mesh topology is a cabling arrangement that you canactually use. A mesh internetwork has multiple paths between two destinations, made possible by the use of redundant routers, as shown in Figure 2.6. Thistopology is very common on large enterprise networks because it enables thenetwork to tolerate numerous possible malfunctions, including router, hub, and cable failures. In most cases, when you see a reference to a mesh topology, this is the application being cited.
Figure 2.6 Internetworks can use a mesh topology to provide redundant paths between networks
The term topology usually refers to the arrangement of cables that forms a network, but it doesn't have to. Although wireless networks use what are calledunbounded media, the computers still have specific patterns they use to communicate with each other. Wireless LANs have two basic topologies, the ad hoctopology and the infrastructure topology. In the ad hoc topology, a group of computers are all equipped with wireless network interface adapters and are able to communicate freely with each other. This provides complete freedom of movement for all of the computers on the network, as long as they remain inside the communication range of the wireless technology. This topology is useful for a home or small business network that consists of only a handful of computers, and for which the installation of cables is inconvenient, impractical, or impossible.
An infrastructure network consists of wireless-equipped computers that communicate with a network using wireless transceivers connected to the LAN by standard cables. These transceivers are called network access points. In this arrangement, the wireless computers do not communicate directly with each other. Instead, they communicate only with the cabled network via the network access points. This topology is better suited to a larger network that has only a few wireless computers, such as laptops belonging to traveling users. These users have no need to communicate with each other; instead, they use the wireless connection to access servers and other resources on the corporate network.
There are three primary types of cable used to build LANs: coaxial, twisted-pair, and fiber optic. Coaxial and twisted-pair cables are copper-based and carry electrical signals, and fiber optic cables use glass or plastic fibers to carry light signals.
Coaxial cable is so named because it contains two conductors within the sheath. Unlike other two-conductor cables, however, coaxial cable has one conductor inside the other, as illustrated in Figure 2.7. At the center of the cable is the copper core that actually carries the electrical signals. The core can be solid copper or braided strands of copper. Surrounding the core is a layer of insulation, and surrounding that is the second conductor, which is typically made of braided copper mesh. This second conductor functions as the cable's ground. Finally, the entire assembly is encased in an insulating sheath made of PVC or Teflon.
The outer sheath—also called a casing—of electrical cables can be made of different types of materials, and the sheath you use should depend on local building codes and the location of the cables in the network's site. Cables that run through a building's air spaces (called plenums) usually must have a sheath made of a material that doesn't generate toxic gases when it burns. Plenum cable costs more than standard PVC-sheathed cable and is somewhat more difficult to install, but it's an important feature that should not be overlooked when you are purchasing cable.
Figure 2.7 Coaxial cable consists of two electrical conductors sharing the same axis, with insulation in between and encased in a protective sheath
There are two types of coaxial cable that have been used in local area networking: RG-8, also known as thick Ethernet, and RG-58, which is known as thin Ethernet. These two cables are similar in construction but differ primarily in thickness (0.405 inches for RG-8 versus 0.195 inches for RG-58) and in the types of connectors they use (N connectors for RG-8 and bayonet-Neill-Concelman [BNC] connectors for RG-58). Both cable types are wired using the bus topology.
Because of their differences in size and flexibility, thick and thin Ethernet cables are installed differently. On a thick Ethernet network, the RG-8 cable usually runs along a floor, and separate AUI cables run from the RG-8 trunk to the network interface adapter in the computer. The RG-58 cable used for thin Ethernet networks is thinner and much more flexible, so it's possible to run it right up to the computer's network interface, where it attaches using a T fitting with a BNC connector to preserve the bus topology.
Run the ThinEthernet video located in the Demos folder on the CD-ROMaccompanying this book for a demonstration of a thin Ethernet computer connection.
Thick Ethernet and thin Ethernet are also known as 10Base5 and 10Base2, respectively. These abbreviations indicate that the networks on which they are used run at 10 Mbps, use baseband transmissions, and are limited to maximum cable segment lengths of 500 and 200 (actually 185) meters, respectively.
Coaxial cable is used today for many applications, most noticeably cable television networks. It has fallen out of favor as a LAN medium due to the bus topology's fault-tolerance problems and the size and relative inflexibility of the cables, which make them difficult to install and maintain.
Twisted-pair cable wired in a star topology, is the most common type of network medium used in LANs today. Most new LANs use UTP cable, but there is also a shielded twisted pair (STP) variety for use in environments more prone to electromagnetic interference. Unshielded twisted pair cable contains eight separate copper conductors, as opposed to the two used in coaxial cable. Each conductor is a separate insulated wire, and the eight wires are arranged in four pairs, twisted at different rates. The twists prevent the signals on the different wire pairs from interfering with each other (called crosstalk) and also provide resistance to outside interference. The four wire pairs are then encased in a single sheath, as shown in Figure 2.8. The connectors used for twisted-pair cables are called RJ45s; they are the same as the RJ11 connectors used on standard telephone cables, except that they have eight electrical contacts instead of four or six.
Figure 2.8 UTP cable has four separate wire pairs, each individually twisted, enclosed in a protective sheath
Twisted-pair cable has been used for telephone installations for decades; its adaptation to LAN use is relatively recent. Twisted-pair cable has replaced coaxial cable in the data networking world because it has several distinct advantages. First, because it contains eight separate wires, the cable is more flexible than the more solidly constructed coaxial cable. This makes it easier to bend, which simplifies installation. The second major advantage is that there are thousands of qualified telephone cable installers who can easily adapt to installing LAN cables as well. In new construction, the same contractor often installs telephone and LAN cables simultaneously.
Unshielded twisted pair cable comes in a variety of different grades, called categories by the Electronics Industry Association (EIA) and the Telecommunications Industry Association (TIA), the combination being referred to as EIA/TIA. These categories are listed in Table 2.1. The two most significant UTP grades for LAN use are Category 3 and Category 5. Category 3 cable was designed for voice-grade telephone networks and eventually came to be used for Ethernet. Category 3 cable is sufficient for 10-Mbps Ethernet networks (where it is called 10Base-T), but it is generally not used for Fast Ethernet (except with special equipment). If you have an existing Category 3 cable installation, you can use it to build a standard Ethernet network, but virtually all new UTP cable installations today use at least Category 5 cable.
Most Ethernet networks use only two of the four wire pairs in the UTP cable, one for transmitting data and one for receiving it. However, this does not mean that you are free to utilize the other two pairs for another application, such as voice telephone traffic. The presence of signals on the other two wire pairs is almost certain to increase the amount of crosstalk on the cable, which could lead to signal damage and data loss.
Table 2.1 EIA/TIA UTP Cable Categories
Voice-grade telephone networks only; not for data transmissions
Voice-grade telephone networks, as well as IBM dumb-terminal connections to mainframe computers
Voice-grade telephone networks, 10-Mbps Ethernet, 4-Mbps Token Ring, 100Base-T4 Fast Ethernet, and 100Base-VG-AnyLAN
16-Mbps Token Ring networks
100Base-TX Fast Ethernet, Synchronous Optical Network (SONET), and Optical Carrier (OC3) Asynchronous Transfer Mode (ATM)
1000Base-T (Gigabit Ethernet) networks
When you install a network with a particular grade of cable, you must be aware of more than the category of the cable. You must also be sure that all of the connectors, wall plates, and patch panels you use are rated for the same category as the cable. A network connection is only as strong as its weakest link.
Category 5 UTP is suitable for 100Base-TX Fast Ethernet networks running at 100 Mbps, as well as for slower protocols. The standard for Category 5e UTP cable was ratified in 1999 and is intended for use on 1000Base-T networks. 1000Base-T is the Gigabit Ethernet standard designed to run on UTP cable with 100-meter segments, making it a suitable upgrade path from Fast Ethernet. The Category 5e standard does not call for an increase in the frequency supported by the cable over that of Category 5 (both are 100 MHz), but it does elevate therequirements for some of the other Category 5 testing parameters and adds other new parameters. In addition to the officially ratified EIA/TIA categories, there are other UTP cable grades available that have not yet been standardized. A series of numbered cable standards (called levels) from Anixter, Inc. is currently being used as the basis for UTP cables that go beyond the performance levels of Category 5e.
There is a Fast Ethernet protocol called 100Base-T4 that is designed to use Category 3 UTP cable and run at 100 Mbps. This is possible because 100Base-T4 uses all four wire pairs in the cable, whereas 100Base-TX uses only two pairs. See Chapter 5, "Data-Link Layer Protocols," for more information.
Shielded twisted pair cable is similar in construction to UTP, except that it has only two pairs of wires and it also has additional foil or mesh shielding around each pair. The additional shielding in STP cable makes it preferable to UTP in installations where electromagnetic interference is a problem, often due to the proximity of electrical equipment. IBM, which developed the Token Ring protocol that originally used them, standardized the various types of STP cable. STP networks use Type 1A cables for longer runs and Type 6A cables for patch cables. Type 1A contains two pairs of 22 gauge solid wires with foil shielding, and Type 6A contains two pairs of 26 gauge stranded wires with foil or mesh shielding. Token Ring STP networks also use large, bulky connectors called IBM data connectors (IDCs). However, most Token Ring LANs today use UTP cable.
Token Ring networks, both UTP and STP, use the ring topology implemented in a MAU, even though the cable is installed in the form of a star.
Fiber optic cable is a completely different type of network medium than twisted-pair or coaxial cable. Instead of carrying signals over copper conductors in the form of electrical voltages, fiber optic cables transmit pulses of light over a glass or plastic filament. Fiber optic cable is completely resistant to the electromagnetic interference that so easily affects copper-based cables. Fiber optic cables are also much less subject to attenuation—the tendency of a signal to weaken as it travels over a cable—than are copper cables. On copper cables, signals weaken to the point of unreadability after 100 to 500 meters (depending on the type of cable). Some fiber optic cables, by contrast, can span distances up to 120 kilometers without excessive signal degradation. Fiber optic cable is thus the medium of choice for installations that span long distances or connect buildings on a campus. Fiber optic cable is also inherently more secure than copper because it is impossible to tap into a fiber optic link without affecting normal communication over that link.
A fiber optic cable, illustrated in Figure 2.9, consists of a clear glass or a clear plastic core that actually carries the light pulses, surrounded by a reflective layer called the cladding. Surrounding the cladding is a plastic spacer layer, a protective layer of woven Kevlar fibers, and an outer sheath.
Figure 2.9 Fiber optic cable has a glass or plastic core surrounded by cladding that reflects the light pulses back and forth along the cable's length
There are two primary types of fiber optic cable, singlemode and multimode, with the thickness of the core and the cladding being the main difference between them. The measurements of these two thicknesses are the primary specifications used to identify each type of cable. Singlemode fiber typically has a core diameter of 8.3 microns, and the thickness of the core and cladding together is 125 microns. This is generally referred to as 8.3/125 singlemode fiber. Most of the multimode fiber used in data networking is rated as 62.5/125.
Singlemode fiber uses a single-wavelength laser as a light source, and as a result, it can carry signals for extremely long distances. For this reason, singlemodefiber is more commonly found in outdoor installations that span long distances, such as telephone and cable television networks. This type of cable is less suited to LAN installations because it is much more expensive than multimode cable and it has a higher bend radius, meaning that it cannot be bent around corners as tightly. Multimode fiber, by contrast, uses a light-emitting diode (LED) as a light source instead of a laser and carries multiple wavelengths. Multimode fiber cannot span distances as long as singlemode, but it bends around corners better and is much cheaper. Fiber optic cables use one of two connectors, the straight tip (ST) connector or the subscriber connector (SC), as shown in Figure 2.10.
Figure 2.10 The ST and SC connectors used with fiber optic cables
Installing fiber optic cable is very different from copper cable installation. The tools and testing equipment required for installation are different, as are thecabling guidelines. Generally speaking, fiber optic cable is more expensive than twisted-pair or coaxial cable in every way, although prices have come down in recent years.
Match the applications in the left column with the network cable types that best suit them in the right column.
| || |
For each of the following scenarios, specify whether the network will function properly based on the information given. If not, explain why.