Installing the cable for your network can be the easiest part of the deployment process or the hardest, depending on the environment where you will be building the network and how concerned you are with its esthetic appearance. You can connect a handful of computers in a single room using prefabricated cable in a manner of minutes, just by running the cables along the walls of the room. It is only when you have to deal with multiple rooms or installations where you want the cables hidden in the walls or ceilings that difficulty arises.

This chapter introduces you to some of the skills required to perform a cable installation that suits your needs and your tastes. Just as important, however, is to know when you are in over your head, so the sections that follow will also identify the circumstances under which you're better off getting professional help.

As part of the network planning and design process, you should have already determined how many computers you’re going to connect and where they're going to be located. On a two-computer network, you can connect the computers tgether directly, nut in the case of a network connecting three or more computers, you're going to need a hub, and you should also decide where you’re going to place it. The only problem that remains is how you are going to get the cables from each computer to that hub.

The difficulties you encounter during the wiring installation process depend largely on the locations of the computers. The following sections examine some of the most common scenarios for small network installations and the complexities involved in each one.

The simplest type of LAN involves two computers, reasonably close to each other, that you want to connect. In this case, a hub is not needed because you can use a single crossover cable to connect the two systems. The computers can be up to 100 meters apart, and the installation of the cable itself is just like any of the other scenarios discussed in the following sections.

The individual wires within a UTP cable have specific functions that are defined by the Ethernet specifications. 10BaseT and 100BaseTX networks only use two pairs of wires, to which the specification assigns transmit data (TD) and receive data (RD) signals in both positive and negative polarity. The remaining two pairs have no function at all. Thus, the wiring diagram for an 8-pin RJ-45 connector uses the values shown in Figure 5-1.

A crossover cable is an exception to general Ethernet practices. For network communications to occur, the signals transmitted out over the TD pins in one computer's NIC must reach the RD pins in the destination computer's NIC. A typical UTP cable is wired straight through, that is, each pin on the RJ-45 connector at one end is wired to the same pin on the connector at the other end (see Figure 5-2).

As the signal travels over the cable, the hub crosses the signals so that the TD pins on each end of the connection are connected to the RD pins at the other end (see Figure 5-3).

If you were to use a single standard cable to connect the NICs on two computers, network communications would not occur because the TD signals on one computer would be transmitted to the TD pins on the other, not the RD pins. To connect two computers directly in this way, you must use a crossover cable that transposes the TD and RD signals internally, replacing the function of the hub.

Crossover cables are not as easily available as straight-through UTP cables; you usually will not find them at the local computer store. However, they are available from many online and mail order sources. For a home or business with only two computers, two NICs, a crossover cable, and Windows 98 are all you need to build a simple network.

If all of the computers that you want to connect are located in one room, you can usually place the hub in any convenient location and run the cables around the room. It's not likely that your one room will require cables longer than 100 meters, so hub placement is not a major issue. Depending on the arrangement of the workstations in the room, you can probably get away with leaving the cables loose and running them along the walls and behind desks. If the room uses cubicles, the units usually have some sort of cable conduit running through the common walls.

One arrangement of this type that can conceivably complicate the situation is if you have desks in the middle of the room that must be connected to other network devices along the walls. In this case, you will have to some special method to get the cable across the open floor while preventing it from being trodden on or damaged in other ways. This method will usually involve the use of a cable housing or protection device or the installation of the cable within a column, wall, or ceiling.

An eternal, or baseboard, installation is a network in which the cables are ostensibly exposed, but protected by securing them to a wall or baseboard, or by encasing them in a protective cover of some sort. This technique may be as simple as stapling the cable along the baseboard to prevent it from being pulled or tripped over, but it can also include a variety of aftermarket products designed to house and protect network cables. When the networked computers are located in two or more rooms, an external installation can go through a simple hole drilled in a wall or run through a doorway.

External installations provide greater flexibility, because you still have access to the cable after the installation, and they enable you to run cable in environments where an internal installation would not be practical, such as an office with cinder block walls. Depending on the products you use to secure and protect the cables, an external installation can use either prefabricated or bulk cable.

An internal installation is the most difficult and expensive method of cabling your network, but it also provides the most protection for the cable and the most professional appearance. The cables are completely hidden in the walls, ceilings, or floors of the rooms, with the connectors installed in a wall plates for each computer and a cabling nexus or patch panel for the hub. You then use short patch cables to connect the computers to the wall plates and the hub ports to the patch panel ports.

Internal installations do not use prefabricated cables, because the cables do not attach to standard male RJ-45 connectors. Instead, the wires within the cable connect directly to the leads in the wall plates and patch panel. Thus, to perform this type of installation, you would purchase the cable in bulk, pull it through the walls and ceilings, and then connect it to the devices at each end. The installation process requires special tools for cutting the cable, punching it down into the patch panel, and testing it. In addition, there may be no small amount of general carpentry and drywall work involved, depending on the complexity of the location.

Clearly, an internal installation is a much more complicated and expensive process than any of the other cabling strategies listed here. You not only have to purchase additional hardware, such as the wall plates, patch panel, and patch cables, but the amount of time you will have to expend is much greater. An internal installation is also permanent; you can't coil up your cables ad move them to a new location if you should have to move your network.

Finally, there is a significant element of uncertainty involved. You never know what you're going to find when you start punching holes in your home or office walls. You may find a fire wall between two rooms that you can't penetrate, ducts and plumbing that interfere with your planned cable path, or any number of other surprises. For all of these reasons, internal installations usually not worth the trouble for a small home or office network. When an internal installation is justified, it is usually a task better left to a professional cable installer unless you are a veteran do-it-yourselfer.

Much of the time, a cable installation will include elements of all of these types. You may have cable running along the baseboards within a given room and then go through a wall or floor to reach another room. The object is to create an installation that satisfies your own performance, economic, and esthetic requirements.

Apart from a few general guidelines, there are no set rules for how you install your network cable. Depending on the nature of the building in which your network will be located, you may come up with an ingenious strategy for installing or hiding the cables. This is perfectly fine, as long as you observe the limitations outlined in the following sentences.

The maximum length for a cable segment in a 10BaseT or 100BaseTX network is 100 meters, but you've seen in Chapter 4, Purchasing Network Components, that there may be other length constraints imposed by the type of cable you purchase. Stranded cables provide added flexibility and are safe to move around, but their increased susceptibility to attenuation limits their length to approximately 20 feet or less.

Most prefabricated cables come in lengths no longer than 50 feet, and this is a good rule of thumb to use when you’re arranging your computers. There are a few manufacturers that make 100-foot prefabricated cables; a reputable maker will only use cable that is suitable for that distance. It's usually not a good idea to push the envelope by using cables that approach the limits defined by the Ethernet specifications. Although you can get 100-meter and longer cables custom made, the problems that can result from pushing or exceeding the maximum allowable length can be intermittent and difficult to diagnose.

Another practice to be avoided wherever possible is the use of special connectors to span a given distance by joining two or more shorter cables together. The additional breaks in the cable can cause the signal to degrade more than it normally would when spanning that distance. In some cases, the use of connectors like these is unavoidable, such as when a PC Card network adapter comes with a short cable that has a proprietary interface to the card. In cases like this, be sure that the connector you use has the same rating (Category 5, for example) as your cable.

On a network where all the computers are located in the same room or in adjacent rooms, the location of the hub is not critical, since all of the cables will be relatively short. However, if you have to span a long distance to connect machines, you might want to plan on placing the hub near the center point between them.

The hub itself should not need attention while the network is operating. Most models have lights that blink as data is being transmitted, and a hub may have another light that indicates when a packet collision occurs, but you won’t need to pay attention to these except in a troubleshooting situation. As a result, you can place the hub on the floor, behind a desk, or in any other hidden location, as long as it has access to an electrical outlet and to the cables coming from all the computers.

Large, elaborate networks with internal cable installations use multiple hubs, daisy-chained together and mounted on racks, with cables connecting the hub ports to the nearby patch panel. This makes it easier to label the ports and cables, so that administrators can identify which ports and cables are associated with a specific computer. On a small network, this type of elaborate organization is not necessary, although there is nothing to stop you from labeling your cables in the same way.

However you decide to install the cabling for your network, you must take pains to see to it that the cable itself is protected from conditions that could damage it. Cable should never be left in an exposed area where people can walk on it or roll chairs and other wheeled objects over it. Thus means that you should either avoid running cable across an open floor by following the walls instead, or use a protective housing that can withstand the weight of people or furniture without compressing the cable.

The cable should also not be pinched or pulled, meaning that if the cable runs through a doorway, secure it so that it cannot get squeezed in the door as it closes and avoid leaving loops of excess cable in places where people can trip over them. Any of these environmental conditions can weaken the cable and eventually damage it.

In many cases, the effect of this type of damage is not immediate. A computer may have intermittent network communication problems for some time as the connection is disturbed before it gives out altogether. This type of situation can be infuriating to the user that can't get his or her work done and frustrating to the administrator that can't reliably duplicate the problem or find an obvious cause. The only recourse is to replace the cable entirely which, depending on the type of installation, can conceivably be a major undertaking.

External cabling can be as easy as running cables along the wall and dropping them behind desks and other furniture. However, there are tools and procedures that you can use to secure the cable from disturbance and protect it from damage. If you will be installing cable to computers in different rooms, you should definitely take steps to anchor the cable to the path you've selected.

The following sections describe the products and strategies that you can use to assemble a functional network without knocking big holes in your walls or buying special tools.

There are a lot of products on the market that you can use to organize the many cables that come out of your computer, some of which are specifically intended for use with network cables. The assumption with these products is that you will be using prefabricated UTP cables and that you want to secure them without installing them inside walls and ceilings. The products that you choose depend largely on how much protection you require for the cable and how important the appearance of the installation is to you.

If, for example, you have to run the cable around a room lined with furniture, then there might not be any need for an elaborate cable management system. On the other hand, in an environment where furniture or other heave objects are frequently moved around the room, you will probably want to protect the cable as much as possible.

The simplest method of anchoring network cables in place is to string them along the walls and staple them into the baseboard. However, while you can conceivable do this, you should take pains to do it properly. The standard staple gun found in the typical homeowner's toolbox is not acceptable for this application. The square-headed staples these guns use will pinch the cable too tightly and could actually break the wires within the sheath.

There are staple guns that use round-headed staples intended for anchoring cable, but these can pinch too tightly as well. If you can't easily move the cable back and forth within the staples, then the installation is too tight. If you do end up using one of these staple guns, be very careful not to pierce the cable sheath as you're stapling it to the wall. If you accidentally do this, then discard the cable and replace it with a new one.

The only type of staples that I would unhesitatingly recommend for cable installations are the type that you insert manually with a hammer. These can be round-headed metal staples with an extra cap at the top to provide a hammering surface, but better still are those with a semicircular plastic holder for the cable and a connected wire brad that goes into the wall. Since only the brad gets hammered, there is little chance of the cable being pinched or damaged, either by the staple itself or by misplaced hammer blows.

Cable ties (see Figure 5-4) are a step above staples, as they can usually hold more than one cable and are sometimes movable. The simplest of these are the locking plastic ties that you can use to bundle cables together, some of which have a ring that you can use to screw or nail them to a wall or floor. Other companies make plastic rings that open and lock closed with an adhesive backing that sticks to a wall. These enable you to anchor the cable in one position and also open so that you can insert additional cables later. You can also improvise your own solution using tape or the plastic ties sometimes supplied with trash bags.

The best cable ties consist of a Velcro strap that you can use to bind several cables together. The strap might be free-standing, or attach to a wall using screws or an adhesive backing. This type of cable tie holds the cables securely but is also very gentle on them. You can easily open the Velcro strap to add or remove cables.

You use cable ties by laying out all of your cables loosely along the walls to which you will secure them, and then wrapping the cables in bundles and sticking or screwing the ties to the wall to anchor the bundles. This method is good for keeping cables away from foot traffic and accidental yanks, but it is still possible for the cables to be damaged when an object bangs into them against the wall.

A raceway is a hollow vinyl or plastic channel, usually rectangular and with a removable top for easy access, that attaches to a wall using screws or adhesive and completely encloses a cable along its length (see Figure 5-5). For an external cable installation, a raceway provides the maximum possible protection for the cable and the neatest appearance. The tube is typically a putty color, and is manufactured in modular form with various fittings available that extend the tube around corners and up and down walls.

Some raceways, such as those made by Belkin Components, are simply open-ended tubes that house one or more prefabricated cables along their length and allow the connectors to protrude from the ends. This type is perfectly adequate for a small network installation. More elaborate products, such as the Black Box Media Track and Hubbell PolyTrak systems, include a wide range of fittings that enable you to run the raceway around corners and up and down walls. Connection boxes are also available that house the RJ-45 connectors for the cable. With this kind of system, you can perform a customized bulk cable installation without penetrating the walls and ceilings of your home or office.

As hard as you might try to avoid it, there are sometimes situations in which it is necessary to run a cable across an open floor. The problem might be a doorway that you can't run cable around or a desk located in the middle of a room. In these cases, it is dangerous to leave a loose cable on the floor, both to the cable itself and to the people walking nearby. Running the cable under the carpet is not a good idea, because the full weight of a person stepping on the carpet above still compresses the cable.

To address this problem, you can buy a type of flexible plastic cable raceway that sits flat on the floor and provides a rigid housing to protect the cable. The raceway itself is curved to form a gentle bump on the floor that chairs can roll over and that people can walk on without tripping over, as shown in Figure 5-6.

A service pole is a hollow post that runs from floor to ceiling, enabling you to run cable through a drop ceiling and then bring it down to desk or floor level anywhere in the room. This can be a good alternative to floor cable covers when you have a cluster of desks in the middle of a room that need access to multiple network cables. The pole will typically have knockouts that enable you to install RJ-45 connectors into the pole itself or leave them as holes through which you can run prefabricated cables. In many cases, the pole also provides electrical outlets.

As you run cables around your home or office, the most common obstruction to your cable path will be doorways, and there are several ways that you can deal with them. If you have to go past a doorway, the best technique is to secure the cable up and over the door with staples. However, be sure to factor in the extra distance involved in your overall cable length. Running from the floor up and over an average home doorway and then down to the floor again requires about 18 additional feet of cable.

Because the cable runs past eye level in an installation like this, it will be much more noticeable to people walking by. It's a good idea to buy cables in a color that approaches your walls or woodwork in order to hide them as much as possible. You can also paint over the cable after it is installed; while the effect may not be suitable for Better Homes and Gardens, it will be better than having your white-painted door surrounded by a black cable.

Another method of cabling past a door that is less noticeable is to run the cable under the threshold in the doorway, if there is one and it is removable. You can also install a threshold if one doesn't exist. This method requires the you be extremely careful during the installation, however. Be sure that the threshold fully covers the cable without pressing down on it and without any sharp edges at the ends that could cut into the sheath.

Both of these methods can be difficult if you have to run multiple cables around the doorway. If this is the case, you could conceivably mount a raceway around the door that holds two or more cables, or you could run a single cable over the door and use an additional hub on the other side to connect the additional cables you need. This is a valid technique for getting around any obstacle with a single cable instead of two or more, as long as you observe the Ethernet cabling guidelines with respect to hub placement. See "Expanding a Network," later in this chapter, for more information on using multiple hubs.

Another cabling problem involving doors is the need to run cable between rooms. Sometimes the traditional solutions used for internal installations, such as running cables inside walls or drop ceilings, are undesirable or impractical. However, it is still possible for a network to connect computers in different rooms without too invasive an installation by using external cabling.

The most obvious solution is to run the cable into the next room in the same way that everything else gets there: through the door. Depending on the door itself and the type of flooring in the rooms, you may be able to run cable around the door jamb near the floor. If there is sufficient room between the floor or carpet and the bottom of the door, the cable should not be pinched when the door closes. Be sure to anchor the cable securely so that it cannot shift or get caught in the door.

If there is not a sufficient gap between the door and the floor to permit the cable to pass without damage, you might want take the door off of the frame and use a sander on one of the lower corners until the cable can pass through easily with the door closed.

No matter how you do it, though, a cable running through a door is going to be noticeable. Another method for running cables between rooms is to go through a wall. Unlike an internal installation, this procedure need not be terribly invasive. Assuming that you're dealing with interior walls made of drywall or plaster and lathe, you should be able to drill a small hole near the baseboard just large enough for the cable to slip through.

The easiest way to do this is to get a drill bit that is as long as your walls are thick (8 to 12 inches, usually) and drill directly through both sides of the wall into the next room. However, it is possible to use a standard drill bit to make a hole on one side, and then use a wire coat hanger or a long, thin screwdriver to poke a locator hole before drilling through from the opposite side. You can also use the coat hanger or a narrow wooden dowel to fish the cable through the wall by taping the cable to one end and feeding the other end through the two holes.

If you use prefabricated cables, you will have to drill larger holes in order to fit the connectors through. For this reason, the best place to put the holes is behind furniture or some other obstruction in both rooms that will keep the cables hidden. You can also get plain wall plates with no sockets mounted in them to cover the holes while enabling you to pull the cable through, or caulk around the cable where it passes through the holes in order to make the installation neater and secure the cable in place.

As with cabling around and through doors, the need to run more than one cable through a wall can be a problem, the best solution for which is to use an additional hub, as long as you conform to the Ethernet cabling guidelines.

An internal cable installation is much more complicated than an external one, not only because the cable runs themselves are more invasive, but because you have to manually connect the wires inside the cable to the wall plates and patch panels at both ends of each connection. If you don’t have the skills to do it yourself, understanding the procedures involved in performing such an installation may persuade you to opt for external cables instead or hire a professional to do the cabling for you. In either case, the sections that follow provide you with cabling information that can help you to perform an external installation yourself or evaluate the capabilities of an outside contractor.

An internal cable installation consists of the following basic steps:

1. Pull a separate length of bulk cable from the location you've chosen for the hub to the site of each wall plate.
2. Connect the wires in the cable to each wall plate and install them into the wall.
3. Connect the wires at the hub ends of the cables to the punch down panel.

This summary makes the process seem quite simple, and in theory it is. However, in practice, the difficulty of actually pulling the cable is dependent on the layout and construction of the site. In addition, there are tools that are either recommended or required for each of the three steps that add significantly to the cost of the installation. Unless you plan on installing other networks in the future, this cost alone make an do-it-yourself internal installation impractical.

The route through your home or office that the cable takes is a crucial element of the installation process. Running cables in the wrong places can make the installation process more difficult, compromise the performance of the network, or violate the building codes in your area. Before you actually begin the installation, you should create a diagram that shows where the patch panel and all of the wall plates will be located, as well as the routes that the cables will take.

In order to diagram your network, you may have to explore inside your walls, drop ceilings, attic, basement, and crawlspace in order to determine what obstacles you may find there. Make sure that the cable can reach all the way to the proposed destination, even if it means poking down into wall cavities with a yardstick and a flashlight. If you are installing the cable in a commercial office building, you should try to obtain blueprints of the building that show where the ventilation, electrical, and other conduits are located.

Consider all of the following elements as you plan the routes that your cables will take:

Fire walls, concrete pilings, ventilation ducts, and steel girders are just some of the barriers that you might encounter within a house or office building that will interfere with a cable installation. Penetrating barriers like this in order to run cables through them might be too difficult to be practical, might compromise the structure of the building, and might violate local fire laws or other building codes.

Other types of barriers may simply be inconvenient. For example, be sure to examine the structure of your building's interior walls before you count on running cable down from the ceiling. Sometimes walls are capped with horizontal studs that, if made wood you can easily drill through. If the studs and the caps are made of metal, however, you might consider surface mounting the cable in raceways instead. Some walls also have horizontal members halfway down, which can prevent you from running internal cables to wall plates down near the floor unless you cut a hole in the wall in order to drill through the stud.

In a private home, do not attempt to cut through any sort of concrete, cinder block, or steel structure without consulting a professional regarding the possible consequences to your actions. In an office building, you should always consult the building manager when a cable installation involves the use of any structures other than the walls and ceilings internal to your office suite or unit. You may also have to have an inspection of the work performed before you can complete the installation.

In most office buildings, a drop ceiling is the most convenient place to run network cables. Unfortunately, electricians know this as well and frequently route their electrical cables there. These cables, plus the fluorescent light fixtures commonly found in offices, are a significant source of electromagnetic interference that can interfere with network communications. When you are planning your cable runs, keep the network cable at least five feet away from all power lines and fluorescent fixtures.

Private homes don't usually have drop ceilings or fluorescent light fixtures, so you might end up having to run the cables through the eaves between the ceiling and the floor above, or through interior walls. In these cases, it might be difficult to observe the five-foot rule, but the wooden joists or other members used to frame the building provide a good source of insulation. You should be safe from interference problems if you avoid running network cable between the same two joists or eaves as an electrical cable.

Fire laws are less of an issue in private homes, but in office buildings they are an important consideration. If your building is inspected and your network cable is found to violate the local code, you will be forced to remove it and reinstall it safely. You may also be subject to a heavy fine for the infraction.

If you plan to run network cable through a plenum (an air space used for building ventilation), then you probably will have to use a cable that has been specifically certified for use in plenums. This type of cable does not give off toxic fumes when it burns. Plenum cable is more a good deal more expensive than standard PVC-sheathed cable, which is why all networks don't use it.

Building codes often require private homes to have barriers between floors in the spaces between vertical joists. This is to prevent the flames and hot gases generated by a fire from spreading too rapidly to the floors above. If you drill through these barriers to run cable from one floor to another, you should stuff the hole around the cable with fireproof insulation, to keep the barrier intact.

Avoid running cables in through unfinished attics, over roofs, or in other areas that are subject to intense heat. Heat can degrade the performance of UTP cable.

Be very careful not to disturb any material that might contain asbestos during a cable installation. In the past, heating pipes in homes were often wrapped in an asbestos-based insulating material that, when left alone, is not dangerous. Asbestos was also used in many different forms in office buildings, such as ceiling tiles, that also should not be disturbed. Attempts to secure cable to pipe wrapping or other materials containing asbestos with staples, cables ties, or other means can dislodge the fibers into the air and cause an extremely serious health hazard.

Unless you are stretching the maximum segment cable length to the limit, you should always leave some slack in each of your cable runs, just in case someone later wants to move one end of the cable to another location or you must move the cable itself to avoid a new obstacle placed in the cable path. Professional cable installers routinely leave a few additional coils of cable for every run in an unobtrusive place, such as inside a wall or drop ceiling.

The process of pulling the cables from the central point (where the patch panel is to be located) to the sites for the wall plates can be sweaty, dusty, and time-consuming, but it is not terribly difficult, as long as you are organized. The most important part of this phase of the installation is to label both ends of the cable, so that you can make the proper connections later.

The type of building in which you are installing the network will be the primary factor in determining where and how you will run the cables. Most office buildings have drop ceilings that are very convenient for this purpose. The cables will typically run up the wall on which you will mount the patch panel, through the space above the drop ceiling, and down the walls at the locations of the wall plates.

Office walls are usually made of drywall, which is easy to cut with a keyhole or drywall saw. Older buildings and some homes might have plaster walls, spread over either a drywall-like product called blueboard or old-fashioned wood lathe. These are much harder to cut and may themselves be sufficient reason not to run the cables within the walls.

Installing cables in homes can be much more difficult, so much so that internal installations within a home are rare, except in new construction. If there is unfinished space above or below the floor on which you will be building your network, such as an attic, basement, or crawlspace, then you can use the same basic techniques as in an office. Run the cable up or down from a hole in the wall, then pull it through the unfinished space and down or up to the location of the wall plate. To run cable below a floor or above a ceiling, you may have to drill through the supporting joists (although you should avoid this practice wherever possible). In most cases, you can do this without weakening the structure of the building, but don't drill your holes near any other holes that may already be there.

Moving laterally within a wall in a home installation is usually not possible, because you would have to drill through the studs that form the framing. Unless you're working with new construction or you want to put a hole in the wall next to each stud in order to insert a drill bit, you're out of luck in this respect. Offices, on the other hand, frequently use metal studs instead of wooden ones that have cutouts for cables, pipes, and other services. However, you'll probably find it easier to run the cable through the ceiling as described earlier, even for short distances.

Running cables between floors is usually easier in a home than an office, because the floor structure in a home is made of wood. You should be able to drill up from a basement or down from an attic into the space behind a wall in the room above or below. In an office building, it might be necessary to drill through a concrete slab between floors, unless there is a raceway between floors that has been built into the building for this purpose. If the raceway is full of electrical cables, it might be necessary to drill your own holes anyway. However, you should never undertake a project of this type without consulting the building manager.

Cabling a large network can require hundreds of separate cable runs, meaning that the installers must work with large bundles of cables and keep track of which cable at the central runs to each wall plate. Even when you are installing a small network, you will save yourself a lot of aggravation if you follow the same procedures.

You should start all of your runs at the central point; place your spool of bulk cable there and pull it off the spool as needed for each run. When you work with bulk cable, you don't have to worry too much about the lengths of the individual runs (as long as they are less that the maximum 100 meters defined by the Ethernet standards) because you can simply pull as much cable as you need from the spool and cut it afterwards.

If you will be running your cable through a drop ceiling, start by removing the ceiling panel directly above the patch panel location and start your runs from there. You can drop the end of the cable down through the wall to the patch panel after you've made the run and cut the cable from the spool.

Begin each run by labeling the cut end of the cable with a code that will identify the wall plate at the other end. You can wrap adhesive labels around the cable to do this or write directly on the cable sheath with a permanent marker. Be sure to place your labels far enough back from the cable end so that you can strip off some of the sheath when wiring the wall plate.

When you get the end of the cable to its destination, leave enough slack to run down inside the walls at both end plus a few extra feet for good measure. Return to the central point, label the other end of the cable and cut it, leaving enough slack there as well. You should never cut an unlabeled cable. Using this technique, you will always be able to identify a cable run from either end.

Getting the cable through the ceiling from one location to another is a lot easier when you have at least two people working together. You can open the ceiling tiles at regular intervals and pass the cable to each other in relays. To extend the intervals between open tiles, professionals use a tool called a telepole. A telepole is a hand-held, telescoping pole with a hook on the end to which you attach the cable. When you extend the pole inside the ceiling, the hook pulls the cable along for a much greater distance than you can reach by hand. Using a telepole, you might only have to open every tenth ceiling tile, instead of every other one.

A telepole is not a necessity, however. Like many of the tools associated with network cable installation, it makes the job easier but is probably not worth the additional expense for a small installation. You can make do with a yardstick, by passing the cable from hand to hand, or by using the coil and throw technique. You will have to move your ladders a few extra times, but for a network with a maximum of ten or twelve cable runs, this should not be too much of a burden.

When pulling the cable through the ceiling, be sure to adhere to your cabling plan and avoid all sources of interference. However, you should not secure the cable inside the ceiling until you complete the run and have wired both ends to the wall plate and the patch panel, to be sure that you have the correct length. Once the ends are fixed, you can go back up into the ceiling and connect the cable to the wires supporting the drop ceiling's support grid every few feet using plastic cable ties. If you have multiple cables running to the same area, bundle them together wherever possible. Do not leave the cable resting loose on the top of the ceiling tiles. You do not want anyone else working in the ceiling to disturb your cables or move them closer to a source of interference.

In many instances, you will have to run two or more cables together at least part of the way to a given location. You can simplify pulling subsequent cable runs along the same path by using a pull string. A pull string is simply a piece of strong twine that you tie to the leading edge of the cable and secure in place with electricians tape. As you run the cable through the ceiling, you unspool the string along with cable all the way to the end.

To run a second cable to the same destination, you tie its leading end to the trailing end of the string and pull it through from destination. If you have three or more cables to pull through, you can attach another pull string to the leading end of the second cable. When you finish pulling the first string through the ceiling, you will have two cable runs and another string to pull for a third.

If the path of the second cable has to divert from that of the first part of the way through, you can use the pull string to run the second cable to that point of diversion, and then continue on from there as described earlier. To facilitate the use of pull strings whenever possible, you should always pull your longest cable runs first.

Once you have a cable run through the ceiling, you have to get the ends down through the walls to the patch panel and the floor plate. After cutting a hole in the wall of the appropriate size, you must run the cable down inside the wall from above. The cavity between the walls may or may not be capped at the top with the same material used for the wall's framing. If the wall is capped with wood, you can drill down through it. If the wall uses metal studs and is capped with the same material, you can either look for a knockout designed to provide access to the wall cavity, or try to cut through the metal with heavy tools (which is not recommended).

Horizontal barriers within the walls can complicate the problem enormously. You can either cut a second hole in the wall in order to drill down through the horizontal stud (assuming it's wood) or compromise by installing the wall plate above the barrier. A simpler solution would be to perform a hybrid installation that uses internal cables within the ceiling, but runs down the walls using a vertical raceway and a surface mounted connection box. By cutting a notch with tin snips in the frame supporting the drop ceiling, you can run the raceway all the way up and keep the cable completely hidden from view.

Once you have a clear path down through the wall to the hole you've made, you can usually fish the cable down from the top and have a partner grab it through the hole with a wire coat hanger. There is also a tool called a fish tape that you can use for this purpose. A fish tape resembles a plumber's snake in that it is a narrow, flat metal or fiberglass band with a hook on the end that is wound onto a reel. Since the band is more rigid than a cable, you can push it up from the hole in the wall to the ceiling, hook the cable, and pull it down. You can also run the band down from the ceiling to the hole and pull a cable up, if you prefer to work from the floor up. This is another tool of convenience for the cabling professional that is generally not needed for a small network installation.

If you are installing your network in a space that has an unfinished basement below it, you can also run the cable down through the floor, through the joists supporting that floor, and up again into the wall cavity at the appropriate destination. The most difficult part about this operation is locating the correct spot to drill up from the basement. You can try to use the standard methods of banging on the floor in the right spot while a partner listens from below or measuring from the nearest outside wall, but these never seem to work out terribly well.

One method that works better is to drill down through the floor where it meets the baseboard, right below the hole for the wall plate, with a 1/8 inch drill bit held at a diagonal pointing into the wall. You can then stick the end of a wire coat hanger through the hole you've made and go down to the basement to find it. Measure 2¼ inches from the coat hanger in the direction it is pointing (towards the upstairs wall) and drill straight up through the floor. You should find your hole right in the center of the wall cavity. Just snake the cable up through this hole, go back upstairs, and grab it through the hole you cut for the wall plate.

The wall plates that you will use to connect computers to the network are available either as part of a box assembly that you mount on the surface or flush with the surface of the wall, or as a plate that you screw directly into the wall (see Figure 5-7). You would usually want to use a surface mounted box when you run the cable to the box within a raceway; when the cable is inside the wall, you can use a box that mounts inside wall with the plate screwed to the box, or the plate alone.

In all of these cases, the wall plate has one or more holes into which you can snap the couplers containing RJ-45 sockets. If you elect to use boxes, you mount them onto (or into) the wall and pull the cable through them. Then, you thread the cable through the wall plate and connect the wires to the coupler. After you have connected the wires to the appropriate pins in the coupler, you snap the whole assembly into the wall plate and screw the plate into either the box or the wall.

The patch panel, also called a punch-down block (see Figure 5-8), is mounted on a wall near the place where you want the hub to be located. This is usually an out of the way spot that is well-protected from potential sources of interference and cable damage and that has an electrical outlet for the hub. On a large network, patch panels are typically located in a wiring closet or data center, but for your small network any convenient location will do.

The patch panel includes a bracket that screws to the wall on top of the hole where your cables will protrude. The other part of the device has holes that accept couplers just like those used in the wall plates. You will connect each of the cables to one of the couplers, placing the colored wires in the same slots as in the wall plate. This ensures each cable is wired straight through.

Organization is again a crucial part of this procedure. You must keep track of which cable is associated with the couplers in the patch panel and the wall plates through the building. Most patch panels include a space where you can label each coupler.

The process of connecting the wires inside a UTP cable to the couplers for the patch panel and wall plate is called punching down. This is the only part of the installation process in which a special tool is essential. A punch-down block tool (shown in Figure 5-9) performs three operations for each of the wires in the cable, which are as follows:

1. It presses the wire down into the coupler between two metal pins that hold it in place.
2. It strips the sheath from the wire to provide an electrical connection with the pins.
3. It cuts off the excess wire protruding past the pins.

The couplers and the punch-down block tool you buy must both be of the same type for them to work together. Most of the products on the market today use 110-style blocks, which defines the type of blade the punch-down block tool needs to cut the wire ends. Some tools contain replaceable blades that you can switch out to work with other types of blocks. The punch-down block tool itself usually costs from $50 to $100, making it the single largest expense incurred by a internal cable installation. Since the tool is utterly useless for any other task, you have to decide whether purchasing one for your small network installation is worth the expense. Finding a friendly network installer who would loan you one would obviously be a big plus.

The other standard that affects this part of the installation procedure is the one that dictates which color wires in the cables should connect to which pins in the couplers. There are two version of the EIA/TIA 568 standard, called 568A and 568B, that define different pin assignments for the RJ-45 connectors used in a network installation. The couplers and patch cables you use on your network should both conform to the same revision of the standard.

The wires inside a UTP cable are twisted into four pairs, colored orange, green, blue, and brown. For each pair, the color of the positive wire is solid and the negative wire has a the same color with a white stripe. Figure 5-10 illustrates how the four pairs are distributed in the RJ-45 connector for each of the two standard revisions.

Most of the networks installed today use the newer 568B wiring standard, and more cables and couplers are available for this standard than the older version. There is no functional difference between the two standards; it is only important that you are consistent in all of the products you buy so that the individual wire signals end up in the right places.

Each coupler for a wall plate or patch panel contains eight pins for the eight wires in two row of four, as shown in Figure 5-11; the pins are numbered and color-coded just like the wires. To make the connections, strip about two inches of sheathing off the cable, untwist the ends of each pair of wires, and lay each of the wires down between the pins of the same color. The beginning of the cable sheath should be no more than 1/8 inch from the coupler, and you should only untwist enough of each wire pair to reach the separate pins. Try to keep the cable centered in the coupler so that the same amount of wire is used to connect to each pin. It is the sheathing and the twists that minimize the effects of electrical interference on the cable, so you do not want to expose any more of the wires than is necessary.

Once the wires are all laid out on the coupler, place the end of the punch-down block tool over one of the wires with the blade on the outside of the coupler and the handle tilted slightly outward. When you press down, the tool will push the wire down between the pins and the blade will cut off the excess wire. Remove the cut-off wire ends and repeat for each of the seven remaining wires.

Once you've punched down all of the wires, you can snap the protective covers onto the coupler and then insert the coupler into the wall plate or patch panel. Then screw the plate into the box or the wall or continue punching down the other connections in the patch panel.

Once you have completed the internal cable installation, you use short patch cables with male RJ-45 connectors to connect the patch panel blocks to the hub ports and the wall plates to your computers. These patch cables are standard straight-through UTP cables that should use the same wiring scheme as your couplers. You can buy prefabricated cables for this purpose or make them yourself using the same bulk cable that went into the internal installation.

To make your own patch cables, you perform roughly the same process as you did when punching down, except that you need make RJ-45 connectors this time instead of female, and a different tool, called a crimper, to connect the jacks to the cable ends. A crimper (see Figure 5-12) looks like a large pair of locking pliers that uses a special set of dies in its jaws to press the two halves of an RJ-45 connector together with the wires of a UTP cable arranged in between them.

As you did when punching down earlier, you strip a small amount of sheathing off of a piece of cable and untwist each pair of wires. Laying out the bottom half of the connector, you arrange the eight wires in the order conforming to the version of the EIA/TIA 568 standard you used earlier. Then you lay the top half if the connector on top and use the connector to squeeze the two halves together.

The bottom line, however, is that the practice of making your own patch cables will not save you very much money on a small network. The crimper and die set will cost you at least $50 and you will probably spend quite a bit of time and ruin quite a few connectors getting the technique down right. What's more, even after a lot of practice, your finished product will never be as sturdy and dependable as a good quality prefabricated cable with a molded boot.

Reading the preceding account of an internal cable installation is likely to leave you saying either "gee, that doesn't look so hard" or "gee, I could never do that." If the latter is the case, you might want to consider hiring a contractor to install your network for you. This would be a particularly good idea if your setting up a new office and you need telephone cable installed as well. The procedures for phone cable installation are virtually identical to those for network cable, with the exception of the guidelines imposed by the Ethernet standards, and many cabling contractors routinely do both.

When seeking a contractor for a small network cable installation, you don’t have to go through the same vetting process that you would for a major enterprise network installation. However, you do want to be sure that the people you hire know what they're doing. Based on the information in the preceding sections, you should quiz them about their practices when it comes to avoiding sources of interference and conforming with building codes. Remember, it is you, not the contractor, who will bear the financial and possibly criminal burden if your installation is found to violate building codes or fire laws.

You should also require that the contractors submit a diagram to you of the precise paths the cable will take through your site and make sure that they adhere to it. Ask them which wiring scheme they plan to use. If they say EIA/TIA 568, ask them "568A or 568B?" and see if they know what you’re talking about.

Expandability is one of the basic features of a LAN; you can add new computers to the network at any time, even if you have used up all of the ports in your hub. By connecting a second hub to your network, you can both add ports for additional computers and extend the maximum distance between workstations. This is because an Ethernet hub also functions as a repeater; it amplifies all incoming electrical signals before transmitting them out through the other ports.

In order to connect a second hub to your network, you must utilize the uplink port included on most of the hubs sold today to connect the new device. Because network cables are wired straight through, standard hub ports provide a crossover circuit that causes the transmitted signals from one computer to be sent to the receive pins on the other computers. If you connect the two hubs together using the standard ports, then a signal transmitted by a computer connected to the first hub would be going through two crossovers circuits on its way to a computer connected to a second hub. The result would be a communications failure because the transmit pins on the two computers are connected, instead of the transmit and receive pins.

When you connect the new hub to the uplink port on the original hub, you bypass the original hub's crossover circuit, resulting in a proper connection between any two machines on the network. Be sure not to connect the two hubs by plugging the cable into both uplink ports. This would bypass both crossover circuits and be the equivalent of a straight through connection between computers connected to different hubs.

Ethernet networks are expandable, but there are guidelines you must observe in order to maintain reliable communications between your computers. Even though Ethernet hubs function as repeaters, just amplifying the signal is not enough to ensure proper communication in all conditions. Ethernet also relies heavily on the amount of time required for a signal to get from one computer to another. Therefore, you cannot chain an unlimited number of hubs together and expect the network to function properly. Eventually, the signal will take too long to reach a destination at the other end of the network an communications will break down.

For this reason, the Ethernet guidelines specify a maximum number of hubs that you can chain together in a single collision domain. A collision domain is a local area network that is cabled so that any two computers can cause a packet collision if they transmit at exactly the same time. When you connect computers together with hubs, they are said to all be in the same collision domain, no matter how many hubs you use.

The guidelines for 10 Mb/sec Ethernet networks differ from those for Fast Ethernet. The specifications for standard Ethernet networks provide a basic guideline known as the 5-4-3 rule that governs the number of repeaters you can use on a LAN. The 5-4-3 rule states that on a single local area network, two computers can be separated by up to five cable segments, connected by four repeaters (hubs), and only three of those five cable segments can be mixing segments.

This means that you can cable multiple hubs together as long as the packets transmitted by one computer on the network do not pass through more than four hubs on their way to any other computer on the network. Figure 5-13 illustrates how you would use five cable segments and four hubs to create the maximum possible distance between two computers on the same LAN. Since each cable segment can be 100 meters longs, you can conceivably have one computer up to 500 meters away from another.

While the Ethernet guidelines limit a LAN to four hubs, there is no limitation on the number of computers that can be connected to each hub. If you plan on expanding your network on a regular basis, you should consider buying hubs with more ports than you currently need, so that you can easily add more computers to the network.

Fast Ethernet guidelines are more restrictive than those for standard Ethernet, because the cable is being pushed to the limit if its bandwidth. As described in the "Fast Ethernet Repeater Types" section in Chapter 4, Purchasing Network Components, there are two types of Fast Ethernet hubs available, called Class I and Class II, that each have their own strengths and weaknesses.

If you have to use different types of Fast Ethernet cable segments, such as 100BaseTX mixed with 100BaseT4, you must use a Class I hub, but you can only use a single hub for the entire collision domain (see Figure 5-14).

Class II hubs can only connect cable segments of the same type. Since there is no translation between media, the timing delays of Class II hubs are shorter and you can have two of them on a single network. The maximum total cable length that you can have between two computers in the same collision domain with two intermediate hubs is 205 meters. Thus, if you use the maximum cable segment length of 100 meters to connect each computer to its hub, the cable connecting the two hubs can be no longer than 5 meters (see Figure 5-15).

You can install multiple hubs on your network for reasons other than a need for additional ports. As mentioned earlier in this chapter, you can add a hub to increase the maximum distance between two computers, or to run a single cable around or through an obstacle instead of running a bundle of cables.

This chapter concentrates on Ethernet networks running over twisted pair cable, but there are times when any cable at all is impractical. Various technologies for wireless networking have been in development for years, but there are finally a wireless networking products on the market that are simple and inexpensive enough for use in home and small business network environments. In addition, there are some products that utilize the existing cabling in a home or office, such as the telephone or power cables, for network data transmissions. These types of networks do not provide anything approaching the performance levels and reliability of even standard 10 Mb/sec Ethernet, but they can provide basic networking services such as file and printer sharing.

There have been wireless networking products on the market for years, but they have usually been too expensive for anything other than specialized applications and not reliable or secure enough for mission-critical data transfers. Wireless networks use any one of several types of electromagnetic airwaves to transmit data between devices. These technologies include:

Infrared

Narrowband

Spread Spectrum

The biggest problem in the development and deployment of wireless LAN technologies has been the lack of a unifying standard. Most of the products on the market are proprietary systems that force you to choose one vendor and stick with them for all of your wireless needs because their products are not interoperable with those of other vendors. Considering that wireless networking is not a technology that has made great inroads in the networking industry as of yet, the question of the chosen company's continued viability in the market in important.

Fortunately, there is a standard in development by the IEEE 802.11 working group that defines several types of wireless physical layer and media access mechanisms. This should result in products from different manufacturers that can work together to provide a practical and reliable wireless LAN solution.

Wireless networking products can either replace or augment the traditional copper cable. An independent wireless LAN consists of a group of computers, all with wireless network interface adapters, that function as peers. An infrastructure wireless LAN uses one or more dedicated hardware devices called access points to provide wireless access to an existing copper-based network.

For the home or small office, it is best to consider independent wireless LAN products that are specifically marketed for use in these environments. This is because wireless technologies that are designed for use in corporate environments can get very complicated and very expensive. There are several companies that are now producing wireless networking solutions for the home or small office, often in complete kits containing everything you need to connect two or three computers. You can then buy additional equipment to support more systems.

Of course, wireless networking is not necessarily the perfect solution for all network users (or everyone would use it). The problems with the technology are the same as they have been since the idea was conceived; manufacturers have made inroads towards addressing them, but wireless networks do not provide the performance of copper-based Ethernet in several key areas.

Chief among these problem areas is speed. Basic wireless networking solutions typically operate at a maximum of 1 to 2 Mb/sec. Manufacturers like to compare these speeds favorably with the 56 Kb/sec rate of the fastest analog modems, but this comparison is not particularly valid. More telling is the fact that wireless networks run at 1/5 to 1/10 the speed of a standard 10BaseT network and 1/50 to 1/100 the speed of a Fast Ethernet network. This makes the technology suitable for basic file and printer sharing (barely) and for sharing an Internet connection (since the Internet connection itself is almost certainly slower than the network), but for networks of more than three or four machines or applications that require large data transfers, it is woefully inadequate.

Range and the accompanying reliability of the network are another concern. The operational range of these wireless products is 150 to 250 feet, depending on the manufacturer. This is more than enough for the average home or office, and wireless technology obviously provides a mobility that cabled systems lack. However, be aware that network performance may degrade as you move systems farther away from each other, and can be affected by environmental conditions such as the composition of the walls and other barriers between the computers, weather conditions, radio interference, and other factors.

Security is a problem that has largely been addressed by the manufacturers of wireless networking systems. Frequency-hopping spread spectrum networks are difficult to penetrate because the communicating systems change frequencies many times each second, using a pattern that a potential intruder would have to duplicate. For a typical home or business network, this is sufficient protection for the transmitted data. For organizations that require more stringent security, there are products that encrypt all data before it is transmitted, although you are not likely to find these targeted at the home and small business markets.

Diamond Multimedia's HomeFree Wireless and SOHOware's CableFREE are two similar product lines that provide an inexpensive wireless solution for the home or small office in which installing cables is impractical. If, for example, you have two or three computers in your home that you want to connect and that are located in different areas of the house, wireless can eliminate the need for unsightly cable or an elaborate internal cable installation.

Like most of the wireless networking products suitable for homes or small offices use, HomeFree and CableFREE use frequency-hopping spread spectrum technology with frequencies in the 2.4 GHz band. The basic products take the form of a network interface cards that you insert into the computer in the usual way and that support the plug and play standard in Windows 98. The only difference is that the cards use a wireless transceiver and antenna in place of a UTP cable. Another product, Webgear's Aviator, uses transceivers that plug into the computers' parallel or USB ports.

Surprisingly, these products are not much more expensive than standard network interface cards (about $100 for a basic ISA card), and future products will enable you to use wireless technology to connect to your existing copper network, via an access point. This could be a great solution for laptops and other portables. Picture a user being able to walk into the office with a laptop containing a wireless PC Card NIC and log on to the network with no cable connection or hardware manipulations of any kind.

Another type of product that is designed to prevent the need for dedicated network cabling uses the power or telephone lines in a home or office to transmit network data. Unlike wireless, which is current used primarily for specific applications on large networks, telephone and power line networking solutions are intended specifically for use in homes and small offices.

Diamond Multimedia has a version of their HomeFree product called HomeFree Phoneline that provides the same 1 Mb/sec throughput as their wireless product for about half the price and uses the same architecture. The hardware involved is a network interface card that plugs directly into a standard phone line. Intel's AnyPoint Home Network product is fundamentally identical in functionality, except that it uses external devices that plug into the computers' parallel ports.

Both of these products enable you to make and receive telephone calls while using the cables for data networking. The idea of using the existing telephone cables for a LAN is ingenious, but the products suffer from the same throughput limitations as home wireless systems. In addition, you must have access to a telephone jack in order to connect a computer to the network. This requirement can sometimes lead to the same cable installation difficulties you would face with a standard Ethernet network.

Another similar of product suffers from no such shortcomings, since it uses the electrical cables in a home or office to transmit data. Passport by Interlogis (http://www.interlogis.com) also uses parallel port connections, but you plug the transceiver units into any electrical outlet, effectively networking your entire home.

Actually, it would be more accurate to say that Passport networks your entire neighborhood, because any other homes that share the same transformer as yours can also theoretically gain access to your systems. The Passport package includes a wizard that provides increased security for your network, but this is clearly not a technology suitable for the transmission of highly sensitive data.

Power line networking is also the slowest of these alternative networking technologies. Running at approximately 100 Kb/sec, it is ten times slower than the typical wireless product, 100 times slower than traditional Ethernet, and 1000 slower than Fast Ethernet.

Overall, the wireless, telephone, and power line products intended for home use are suitable only for the most basic networking tasks. If you have two or three computers in your home that you want to connect with the least invasive hardware installation and you need only to share a printer, an Internet connection, and perhaps a few files, then this type of network may be right for you. Of the three, the telephone line networking products are likely to be the most reliable in a typical home environment. The power line concept is too slow and the wireless networks are too susceptible to minute changes in the position of the units and environmental conditions.

Whichever type of product you select, it is a good idea to see whether they confirm to any existing standard for the technology, so that you might continue to find support and new products for your network in the event that your original vendor ceases production. The standard for wireless networking products is the IEEE 802.11 specification mentioned earlier, which is supported by the SOHOware products, while a group called the Home Phoneline Network Alliance publishes a standard for telephone line networking products that is supported by Diamond Multimedia.