Networking

Setting up a server room, part 3: Running the cables

Dallas Releford continues his series on setting up a server room with a discussion of cabling. He describes the different types of cables and offers advice on which type might be right for you.


Different cables are applicable in different situations and conditions, and some cables are more effective than others. For instance, fiber-optic cable is very expensive, but it’s so fast that it’s very cost effective if you can use it as the backbone or trunk of your network. Generally, it takes just as much time and effort to run standard cables, and they are much slower. In an Ethernet network, which normally uses thin Ethernet or twisted-pair cable, you can speed things up by using optical fiber. You also can use fiber-optic cables between buildings or floors. The only problem is figuring out which type of cable is best for you.

Types of cable
There are four basic types of cables. Some types are suitable for only certain kinds of installations; others can be used for most installations. Some types require little maintenance; others need ongoing attention.
  1. 1.      Coaxialcable: The two varieties of coaxial cable are thin and thick. Thick coaxial cable isn’t used much anymore—except in certain applications. Thin coaxial cabling, which is mostly used for networks, is usually referred to as thin Ethernet. In general, coaxial cabling isn’t as popular as it used to be because it requires lots of maintenance, it breaks easily, and it’s bulky and hard to use. It’s one of the least cost-effective selections that you can make.
  2. 2.      Twisted-paircable: Shielded twisted-pair (STP) and unshielded twisted-pair (UTP) are the two varieties of twisted-pair cable. They are the most common cabling systems because they’re easy to install and maintain. Twisted-pair cable doesn't break as easily as coaxial cable, and it can run around corners and through walls. Since this type of cabling is so easy to install and maintain, it will save you time and money.
  3. 3.      Fiber-opticcable: This type of cabling system has a conductor that’s made from a special kind of glass or plastic. Whereas other types of cable transmit electrical signals, fiber-optic cable uses laser light as its carrier medium. Fiber-optic cable is the fastest type of cable, but it’s still very expensive and very prone to damage. Furthermore, the connections are so complex that only highly trained technicians should install this type of cabling.
  4. 4.      IBM cable: Created by IBM for IBM Token Ring networks, this type of cable is actually twisted-pair cabling that was designed to meet IBM’s specifications.
Don't confuse coaxial cable with UTP cable. They are slightly different, and they have different properties and specifications. Coaxial cable breaks, and it’s difficult to handle. UTP cable is easier to deal with. It bends easily, and it uses a different kind of connector. UTP cable looks like telephone cabling (and it’s often used for that purpose). Coaxial cable is used for televisions, VCRs and DVD players, and CBs—as well as for computer networking applications.
Perhaps the most widely used cable is twisted-pair cable, which has two conductors wrapped around each other. This cable keeps crosstalk to a minimum, and it’s very easy to use. It’s also cost effective, and it works very well. Of the two types of twisted-pair cable, STP is used most often in Arcnet and Token Ring networks, while UTP is used primarily in Ethernet and Arcnet network topologies.

Packaging
Cables are packaged in different ways. Coaxial cable normally has only one center conductor, a layer of insulation, a layer of (foil shield) aluminum, and fine wire (braid shield) wound around the outer aluminum layer. Then, of course, there’s usually a jacket. This setup is referred to as simplex cable. Duplex cables usually have two conductors or fibers. That way, one fiber transmits in one direction, and the other fiber transmits in the opposite direction. Multifiber cabling refers to cabling that has a single jacket but that has more than one fiber as a conductor. Some fiber-optic cables have thousands of these fibers, but most electrical cables have only a few dozen conductors.

Cable specifications
All cables have several things in common. First, each cable must have either a copper conductor or a glass tube through which electrical or light signals are transmitted. To keep the signal flowing and to limit interference, each cable must be insulated. This insulation also helps control moisture, heat, and other environmental factors. Each cable must have an outer shell or sheath (jacket), which encases and protects the inner components. The jacket keeps the conductor and insulation packed together, and it protects the cable’s insides from water, pressure, stress, and other outside damage.

Of course, insulation will vary from cable to cable. Insulation usually consists of a dielectric material, which doesn’t conduct electricity. Whereas conductors have properties that allow electrons to flow through them, insulation is nonconductive and resists the flow of free electrons. The insulation for fiber-optic cabling is called cladding, which is made of material that has a lower refraction index than the core material of the conductor. The main conductor (the glass laser tube) has a high refraction index; it allows a great deal of light to pass through. The refraction index is a method of measuring how well certain materials reflect light.

If you don’t run your cable through a conduit, then you should use plenum cabling. Plenum cabling is made from fire-resistant materials, and it’s very safe in server rooms and other locations where fires can occur. Most fire codes require you to use plenum cable if a conduit doesn’t cover the cable itself inside the plenum area—the area between the dropped ceiling and the floor above. In the server room, you should use both plenum cable and a rigid conduit. That way, you’ll be doubly protected.

Installation considerations
When selecting and installing cables, you must take several factors into consideration, including attenuation and crosstalk. Attenuation, which refers to the decrease in signal strength of a cable as the cable length increases, is measured in decibels per 100 feet of cable. Repeaters can boost signal strength, but you have to make sure that you place repeaters at frequent enough intervals in order to handle this problem. The workstations or nodes on the network also boost the signal and keep it clean.

Crosstalk is interference that occurs when one signal interferes with another signal. In its simplest form, the signal in one cable tries to talk to the signal in a nearby cable. You can avoid this problem to some extent by keeping cables far away from other cables, electric motors, telephone and electric wires, and fluorescent lights. Of course, the easiest ways to cut down on crosstalk is to run network cabling inside of conduits in high-risk areas. If possible, don’t run two or more network cables side by side.

Finally, remember that coaxial cables use metallic connectors that can pick up stray interference. If they come into contact with anything that carries electricity, you may have a problem. Coaxial cables also tend to pick up electromagnetic interference (EMF) from such electrical sources as generators and electric motors. You shouldn’t use coaxial cabling for large computer systems; you won’t want to use coaxial cabling for those extensive cable runs. UTP cable is a much better choice for most installations.

Coaxial cable connectors
In a typical network that uses coaxial cable, the network is formed by one single run of cable (which is referred to as the backbone or backplane of the network) and terminator connectors on each end of the installation. One end of the cable must be grounded. Each workstation or node of the network connects directly to this cable through a T-connector.

If you’re using thin Ethernet (RG-58) cabling, T-connectors will connect the node, server, or workstation directly to the backplane. Each T-connector must have a butt or terminator connector, and one end must be grounded. Use a 50-ohm terminating resistor for this purpose. Then, use bayonet nut connectors to splice (BNCs) the two cables together. (You can use BNCs to connect two components or devices together, too.) I always wrap several layers of black electrical tape around these connectors—just to make sure that they’re protected from dirt and liquids. This setup is the simplest type of network, and it’s very affordable for small business and private computer users. It’s also one of the easiest to maintain.
You can’t use drop cables with thin Ethernet networks, but you can with thick Ethernet. Thin Ethernet cable has a 3/16-inch diameter, and it’s known as 10Base2. Thick Ethernet cable has a 3/8-inch diameter, and it’s known as 10Base5. However, thick Ethernet cabling schemes aren't used very much nowadays.
Conclusion
Whether it’s fiber-optic, coaxial, twisted-pair, or IBM, cabling can make or break a network installation. Knowing which type of cable to use in an installation will save you a great deal of time and hassle down the road. I trust that this quick introduction to the fundamentals of network cabling will aid you when you install your cables.

Dallas G. Releford has worked in the computer field as a programmer, an MIS manager, and a PC specialist. He has written a novel, which was published on the Internet and which led him to an interest in the electronic publishing field. He writes articles, electronic books, and just about anything else that involves the written word. To learn more about Dallas’ business, visit his Web site, which is called The Editor’s Eye .

The authors and editors have taken care in preparation of the content contained herein, but make no expressed or implied warranty of any kind and assume no responsibility for errors or omissions. No liability is assumed for any damages. Always have a verified backup before making any changes.

Editor's Picks

Free Newsletters, In your Inbox