|
Appendix D
Networking Technologies
As the number of Windows NT administrators increases in direct proportion to the number of Windows NT installations, these new administrators might not have an appropriate background. If these administrators don’t understand NT’s basic networking concepts, they can be at a disadvantage when it comes to supporting Windows NT. This appendix is aimed at providing a grounding in networking technology, and the information contained in it will stand you in good stead, regardless of which future network operating system you support.
LANs And WANs
LANs (Local Area Networks) and WANs (Wide Area Networks) are two of the most common network acronyms. The widespread use of networking means that most organizations now have a mixture of both.
A LAN can be defined as the network you are locally attached to, whereas the WAN provides connectivity between remote locations, for example, connecting the Boston branch office to the New York head office. The line defining a LAN and a WAN often becomes blurred—in fact, the term ELAN is sometimes used to describe an Extended Local Area Network.
Ethernet
The development of Ethernet can be traced back to the 1970s, when a radio-wave-based network named ALOHANET was developed by the University of Hawaii for inter-island data communications. From this basic network concept, the Xerox Corporation developed the Ethernet network technology to run over coaxial cable. Xerox later was joined in the development of Ethernet by Intel and Digital Equipment Corporation (DEC).
Networking computers with Ethernet has become relatively inexpensive; however, in large installations, the actual installation costs of the coaxial cable are the major expense, especially in old buildings that were never designed to take network cables. There are three types of coaxial cable that can be used with Ethernet networks: thick wire, thin wire, and unshielded twisted pair (UTP). Ethernet networks can, and often do, consist of a mixture of all three cable types.
The thick wire coaxial cable is the oldest form of cabling used to connect computers together via Ethernet, and it is both difficult to install and expensive. Attaching new nodes to a thick wire segment involves drilling through the shielding and attaching a transceiver to the cable. Care has to be taken when performing this operation, as causing the shielding of the coaxial cable to touch the inner core of the cable will halt all network traffic. Thick wire coaxial is very robust and can be used in segment lengths of up to 500 meters without any additional hardware.
Thin wire coaxial cable is both considerably cheaper and easier to install than thick wire. However, it is not as robust as thick wire cable and is often damaged—especially when offices are being refurbished. The attachment of nodes to thin wire is achieved by attaching “T” pieces to the cable, which then connect to your computer. However, if there isn’t a free “T” piece available on your thin wire segment, an additional piece of cable will need to be added with an additional “T” piece. The network traffic will also need to be interrupted when the “T” piece is added. (Buildings can be wired with special thin wire connection plates, known as make-before-break connectors. These allow computers to be added to the thin wire segment without interrupting the network traffic.) Thin wire segments can be up 185 meters in length without any additional hardware being required.
Both thick and thin wire coaxial cables need to be terminated at both ends of the segment with Ethernet terminators. The terminators place a 50 ohms impedance at the end of the cable. Because these terminators are connected in a parallel fashion, if the impedance is measured on the segment using a multi-meter, the impedance will be 25 ohms. This is a quick and easy check to verify that a segment is terminated properly. If a terminator is removed for the segment or if the coaxial cable is cut, all network traffic on the segment will be interrupted—this could put a whole department out of action.
The third type of cabling that can be used is unshielded twisted pair (UTP), which provides point-to-point network connections combined into a single network via a network device known as a hub. As the cables are connected point-to-point, damaging a single cable will only affect the network node attached to it. In other words, a single user will be out of action instead of a whole department. New buildings are often cabled with UTP during the construction stage, using a technique known as flood wiring. This technique calls for UTP cables to be installed in every location within the building—these can then be used as required. Unshielded Twisted Pair segments can be up to 100 meters in length.
CSMA/CD
Carrier Sense Multi Access, Collision Detect (CSMA/CD) is the technique used with Ethernet networks to transmit the data onto the network. Only one node may transmit at a time. When a network node wants to transmit data onto the network, that node will first “listen” to the network to see if another node is currently transmitting. This is the Carrier Sense. Multi Access means all network nodes are equal and have the same rights to access the network. Collision Detect is used to detect when there are two network nodes transmitting at the same time. For example, if two network nodes both want to send data across the network, they first check to see if the network is free (no one else is currently transmitting data). If they decide the network is free, they both may send their data at the same time—this would cause a network collision. Both network nodes would recognize that a collision has occurred and will top transmitting. They will both then wait a random amount of time before attempting to transmit the data again. Collisions are expected to occur in an Ethernet network and, as long as the number of collisions is not excessive, they won’t cause a problem.
Ethernet network nodes are identified by unique hardware addresses assigned to the network adapters. Ethernet data is sent across the network in data packets. These packets contain both the destination address and source address, the actual data, and a data check sum.
Bridges, Routers, And Repeaters
As we have already discussed, each type of cable used with Ethernet networks has different characteristics. For example, the maximum thin wire single segment length is 185 meters. To extend your network over this length, additional hardware is required. Bridges, routers, and repeaters can all be used to extend you network.
Repeaters can be used to extend a network by regenerating the data packets. The repeater functions at the lowest level and does not understand the packets it is regenerating. All data is passed through a repeater—no filtering is performed.
Bridges can be used to not only extend network segments, but also to filter network traffic, so only the packets needed on a particular network segment will be passed by the bridge, thereby reducing the overall network load.
Routers are used to logically split networks into different areas. Data packets are directed by the routers across the network to reach their destinations. Not all network protocols can be routed. For example, NetBEUI is a nonrouteable protocol. Routers are not always separate hardware devices—sometimes a host computer will act as a router to route packets through two separate network adapters. More details on network protocols are given in Chapter 7.
Token Ring
Token ring networks are an alternative to Ethernet technology that have been developed by IBM. Token rings use what is called a twin ring topology, which has the advantage that if the primary network cable is broken, the network can automatically be switched over to the fallback ring, so no network disruption will occur.
Token ring network nodes are only allowed to transmit data on the network when they hold the token, the token being a data packet that is passed from node to node on the network. This means that only one node at a time will be transmitting, so no collisions will be created.
Like Ethernet, several different types of cable can be used to construct a token ring network, and each cable type has a different maximum length.
ATM, FDDI, And Fast Ethernet
As networks have increased in size and the applications that run across them have become more complex, the available network bandwidth has become a performance bottleneck. To counter this problem, new network technologies have been developed—namely ATM, FDDI, and Fast Ethernet—that increase the speed of the network from the 10 MB per second for Ethernet and the 16 MB per second for Token Ring up to 100 MB per second.
|
FEEDBACK
