by Louis Masters
This chapter starts with an analysis of the Open Systems Interconnect (OSI) Reference Model and how it relates to data networking. All seven layers (application, presentation, session, transport, network, data link, and physical) are explained, as well as the interaction between the layers. The OSI model gives you a way to visualize the interaction between the many parts of a data transmission.
This chapter concentrates on the physical layer of the OSI model. It covers what types of transmission media are available, gives a comprehensive explanation of how they work, and also provides specific benefits and penalties to using them. A precise comparison table is provided to aid in media analysis. The table is intended to give you an in-depth analysis of most of the better-known physical network transmission methods. This chapter arms you with a healthy knowledge of what certain types of networking media can and cannot do.
In 1983, the International Standards Organization (ISO) created the OSI, or X.200, model. It is a multilayered model for facilitating the transfer of information on a network. The OSI model is made up of seven layers, with each layer providing a distinct network service. By segmenting the tasks that each layer performs, it is possible to change one of the layers with little or no impact on the others. For example, you can now change your network configuration without having to change your application or your presentation layer. The basic OSI model is depicted in Figure 2.1.
The OSI model was specifically made for connecting open systems. These systems are designed to be open for communication with almost any other system. The model was made to break down each functional layer so that overall design complexity could be lessened. The model was constructed with several precepts in mind: 1) Each layer performs a separate function; 2) The model and its levels should be internationally portable; and 3) The number of layers should be architecturally needed, but not unwieldy.
Each layer of the model has a distinct function and purpose:
FIGURE 2.1. The basic OSI model.
Data travels from the application layer of the sender, down through the levels, across the nodes of the network service, and up through the levels of the receiver. Not all of the levels for all types of data are needed--certain transmissions might not be valid at a certain level of the model. A sender-receiver OSI example is shown in Figure 2.2.
FIGURE 2.2. An example of an OSI send/receive.
To keep track of the transmission, each layer "wraps" the preceding layer's data and header with its own header. A small chunk of data will be transmitted with multiple layers attached to it. On the receiving end, each layer strips off the header that corresponds to its respective level. Figure 2.3 illustrates how the data is wrapped by the OSI layers.
FIGURE 2.3. The OSI level wrapper.
The OSI model should be used as a guide for how data is transmitted over the network. It is an abstract representation of the data pathway and should be treated as such.
For Local Area Networks (LANs), there are three principal connection schemes: twisted pair, coaxial, and fiber optic cable. Satellites, lasers, microwave, and the like can also be used for transmitting network information, but a discussion of those technologies is beyond the scope of this book.
Twisted pair (TP) is the most common form of transmission medium in use today. Quite simply, TP is a pair of wires twisted together and combined to form a cable. The entire cable is usually surrounded with a tough PVC sheath to protect it from handling or its environment. Figure 2.4 depicts TP.
FIGURE 2.4. Twisted-pair cable.
TP is normally used to carry data at speeds from 10Mbps to 100Mbps, but the speed can be decreased by a number of error characteristics: data loss, crosstalk coupling, and electromagnetic interference (EMI).
Shielding (screened twisted-pair cable and foil twisted-pair cable) may be added to TP to confine the wires' electric and magnetic fields. But, when you shield TP, you also increase attenuation. Attenuation is the decrease in signal strength from one point to another on the network. The shielding of the cable also causes the resistance, capacitance, and inductance to change in such a way that you may lose data on the line. This loss can make shielded TP undesirable as a reliable transportation medium. Both unshielded and shielded TP can be used in the several hundred meters segment range.
The five major categories of TP cable are based on specifications designed by the Electronic Industries Association and the Telecommunications Industries Association (EIA/TIA). Please note that the EIA/TIA used only unshielded twisted pair (UTP) when it defined the standard wiring categories for twisted-pair cables.
Here are some points to remember:
Coaxial cable, named from the two cable axes that run the length of the wire, is a versatile and useful transmission medium. The cable consists of a solid or braided outer conductor surrounding either a solid or a stranded inner conductor. The conductors are usually separated by a dielectric material, and the entire wire is covered with an insulating jacket. Coaxial wire allows for greater shielding from interference and greater segment distances. Coaxial 10base-5/2 has a transmission rate of 10Mbps. 10base-5 has a maximum segment length of about 500m/segment, whereas 10base-2 is about 180m/segment. Figures 2.5 and 2.6 show a breakdown of coaxial cable.
FIGURE 2.5. Cross section of coaxial cable.
FIGURE 2.6. Side view of coaxial cable.
As the diameter of the coaxial cable increases, the data pipeline increases, and so does the transfer rate. But larger wires are also expensive and require special installation tools, thereby making the installation of large-circumference coaxial cable cost-prohibitive.
Here are some points to remember about coaxial cable:
Optical fiber is a thin, flexible medium that carries data in the form of light waves through a glass "wire" or cable. This transfer medium works for distances exceeding the 1-kilometer range and is extremely secure (there is no electric signal to tap). Fiber optic cables come in two varieties: single mode and multimode (graded). The difference between the two will be explained shortly.
The composition of fiber optic cable is similar to that of the coaxial cable. It has a solid core made up of one strand of ultra-thin glass or sheathed in a plastic covering (cladding), which reflects the light back into the cable's core. The cladding is covered by a concentric layer of thin plastic (jacket) that protects it. When there is more than one fiber, the fibers are grouped together into a bundle and covered with another thin layer of plastic. Figure 2.7 shows a diagram of fiber optic cable.
FIGURE 2.7. Fiber optic cable.
As you know, the 0s and 1s your computer uses are merely on and off states. Non-fiber cable transmits this binary-state data by using pulses of electricity. Fiber, utilizing the same principle, uses light pulses to transmit data. A light source sends the data down the fiber "pipe" where, at the terminating point, it is received and converted back into data that the receiving device can use (see Figure 2.8).
FIGURE 2.8. How fiber works.
If you have a thin piece of fiber wire (glass fiber), light will travel down along the axis of the wire. This is known as light traveling along an axial path, which is what happens in single-mode fiber cable (see Figure 2.9).
FIGURE 2.9. A thin fiber showing a single mode axial path.
However, the power of this type of transmission is extremely limited. To lessen this limitation, wider cables are implemented. With wider cables, however, you encounter the problem that some of the light waves will enter the pipe at different angles and will travel non-axially down the cable (bouncing from wall to wall). These non-axial waves will travel for a greater distance than the waves that travel axially, causing the light to arrive at the terminating point at different times. This is known as modal dispersion (see Figure 2.10).
FIGURE 2.10. A thick fiber showing non-axial paths and modal dispersion.
As the number of modes of light down the pipe increases, bandwidth tends to decrease. In addition to different pulses reaching the destination at slightly different times, too much dispersion also results in light pulses overlapping and "confusing" the receiving end. This results in an overall lower bandwidth. Single mode provides only enough of a data pipe for a single mode of light to be transmitted. This results in speeds greater than tens of gigabits per second and can even support multiple gigabit channels by using different wavelengths of light simultaneously. Thus, multimode fiber is slower than single mode.
The easiest way to decrease dispersion is to grade the fiber cable. Grading synchronizes the faster and slower light paths so that dispersion at the receiving end is limited. Dispersion can also be lessened by limiting the number of wavelengths of light. Both methods lessen dispersion somewhat but still don't approach the speeds reached by single-mode fiber.
The most popular type of multimode fiber in the United States is 62.5/125. The "62.5" is the diameter of the core, and the "125" is the diameter of the cladding (all in microns). Single-mode fiber is most common in the 5-10/125 micron range. Fiber bandwidth is usually given in MHz-km. The relationship between fiber bandwidth and distance is elastic--as distance increases, bandwidth decreases (and vice versa). At 100 meters, multimode usually reaches about 1600MHz at 850nm. Single-mode fiber reaches about 888GHz for that same 100-meter run.
The light source of fiber cable may be either a light-emitting diode (LED) or an injection laser diode (ILD). Single-mode fiber generally uses LEDs as the light-generating device, whereas multimode uses ILDs.
Here are some points to remember about fiber cable:
Table 2.1 shows the advantages and disadvantages of twisted pair, coaxial, and fiber cable. You should weigh each of the advantages and disadvantages relative to the project you have at hand.
| Medium | Advantages | Disadvantages |
| TP | Low cost, easy to install | Unsecure, worst noise immunity |
| COAX | Relatively fast on short runs | Unsecure, poor noise immunity |
| Fiber | Voice, data and video, fast, long distance | difficult to install, limited to point-to-point, expensive |
The choices in the field of high-performance networking have never been as varied as they are today. Newer technologies such as Fast Ethernet, Fibre Channel, FDDI, and ATM have begun replacing their older networking forefathers, such as Token Ring and Ethernet. The following section gives you a sampling of most of the major LAN technology types.
Simple Ethernet is one of the oldest, simplest, easiest, and cheapest LAN technologies to implement. Several varieties are available, based on medium type:
The architecture for all four types is basically the same. They transmit data on a LAN at speeds up to 10Mbps. They all use the CSMA/CD protocol (see the following section) to send data on the network. Currently, the most popular type of this form of Ethernet is the twisted pair variety.
The heart of the Ethernet technology lies in its Carrier Sense, Multiple Access, Collision Detect (CSMA/CD) protocol. Carrier Sense means that each station will check to see if any other station on the network is transmitting. If so, the station will not detect "carrier" and will not transmit. The station will keep trying to "capture the carrier" until the network becomes idle and the carrier becomes available. Collision Detect means that if two stations transmit at the same time, and their signals collide, they will cease their transmissions and try again at a later time (randomly determined). Multiple Access simply indicates that every station is connected by a single line on the network.
Token Ring is an older LAN technology based on a ring architecture. A control station creates a special entity on the network called a token and passes it around the network ring. This token controls which station has the right to transmit on the network. When the token reaches a station on the network that has something to transmit, it "captures" the token and changes the status of the token to busy. It also adds to the token whatever information it wants to transmit and passes it on. It "circulates" through the network until it reaches the station the information was intended for. The receiving station takes the information and passes the token on. When the token reaches the originating station (the station that originally assigned information to it), it is removed from the network and a new token is issued. The cycle begins anew. Figure 2.11 shows the structure of a Token Ring network.
FIGURE 2.11. A Token Ring network.
Token Rings are an orderly and efficient network architecture. There are currently two versions available, one running at 4Mbps and one running at 16Mbps.
Several varieties of Fast Ethernet are available, based on medium type:
Thought of as a high-speed cousin to the older 10base-X Ethernet, 100base-XX is capable of transmitting data over a network at up to 100Mbps.
100VG(Voice Grade)-AnyLAN is another version of the 100Mbps network specification. The main difference is that it uses the Demand Priority Access Method (specification 802.12) in place of CSMA/CD to communicate across the network. 100VG-AnyLAN utilizes media types of CAT 3, 4, 5-UTP, 2-STP, in addition to fiber optics. It also supports the 802.3 and 802.5 formats. This support allows for a smooth transition from previous network topologies. Although 100VG runs at almost 100Mbps on copper wire due to its Demand Priority Access method and its use of tiered repeaters and hubs to help network traffic, it has not yet been certified to run on fiber. This makes it a poor choice for running a 100VG network over a long distance (greater than 100m). Also, most of the equipment available for 100VG is from only a few vendors, making this technology proprietary (at least for the time being). Figure 2.12 shows the tiered hub architecture of a VG network. The tiers or branches enable a better distribution of nodes on the network. The topmost, or parent, hub can have zero or more children. Each child can itself have zero or more children and one parent hub. Each hub contains all of the medium rules for network access (not distributed throughout all of the children). Data travels upward throughout the hubs and into the network.
The Demand Priority Access method is VG's answer to the CSMA/CD protocol. The client or requester now requests access to the network media for the purpose of transmitting information. The server or granter processes the request and sends a signal back to the client when and if the media is ready to use. At this point, the client has control over the media and may transmit its data. Figure 2.13 shows a diagram of this method.
IsoEthernet is unique in that it not only supports the standard 10Mbps Ethernet, but also 96 ISDN B-channels operating at 6.144Mbps, one 64Kbps ISDN D-channel for signaling, and a 96Kbps M-channel for maintenance. The 10Mbps Ethernet channel is used for data packets, and the ISDN B-channel is used for video and audio. IsoEthernet works on existing CAT3 LAN lines and requires no costly cabling upgrades to existing networks.
FIGURE 2.12. VG LAN using tiered hubs.
FIGURE 2.13. VG request and grant.
FDDI, or Fiber Distributed Data Interface, is a stable fiber-based transmission medium capable of speeds up to 100Mbps. It is frequently used as a backbone to large networks, as well as an interim network for connecting LANs to high-speed computers. FDDI is based on a Token Ring topology, but instead of a main single ring to transmit information, it uses two. The first ring is usually the primary ring, and the second is held as a backup. The rings run counter to each other so as to lessen network errors. Future enhancements see the second ring also being used for data transmission, effectively doubling the network transmission rate. Figure 2.14 shows an example.
FIGURE 2.14. An FDDI dual ring architecture example.
CDDI, or Copper Distributed Data Interface, was created mainly as an answer to the high cost of fiber optic cable. There is also the need (for better transmission rates)/want (of faster response time) to use existing shielded and unshielded network cables.
Fibre Channel (FC) is a new intelligent connection scheme that supports not only its own protocol, but also those of FDDI, SCSI, IP, and several others. This will serve to create a single standard for networking, storage, and general data transfer. Originally meant for WANs, FC can easily be converted to LAN standards by using a switch on the network. FC supports both channel and networking interfaces over one computer port, lessening the network burden on the station. It also supports both electrical and optical media on the network, with speeds ranging from 133Mbps to 1062Mbps. A key piece of FC is the fabric--an abstract entity that represents the interim network device, be it a loop, active hub, or circuit switch. FC, for all intents and purposes, is still in the planning stages.
ATM, or Asynchronous Transfer Mode, is the proposed communication standard for broadband ISDN. ATM is a very high performance solution for both local area and wide area networks. ATM makes use of a special high-speed switch that connects to computers by optical fibers (1 for send and 1 for receive). ATM also supports simultaneous transmission of voice, data, and video over one network technology. It is currently available at a speed of 25Mbps, but was originally designed to run at 155Mbps. Future expansion could be in the gigabit or even terabit range. ATM is covered in more detail in Chapter 18, "ATM."
Current Ethernet networks are available in either 10Mbps or 100Mbps size. Gigabit networking increases that bandwidth tenfold, allowing speeds of up to 1000Mbps. Existing Ethernet and Fast Ethernet networks are 100% compatible and easily upgradable to the new gigabit networking architecture. This new architecture supports the CSMA/CD protocol and will be available on fiber, coaxial cable, and even UTP.
This chapter explained the fundamentals of the OSI Reference Model and what it means to a network. Each level of the model was covered along with how they tie together. You reviewed the advantages, disadvantages, and ratings of twisted pair, coaxial, and fiber cable. You learned about older technologies like Token Ring, some common ones like Simple Ethernet (10base-T and so on) and even some fairly uncommon types like Gigabit Ethernet. The chapter ends with a quick reference chart (see Table 2.2) outlining all of the technologies covered, their estimated speeds, and their maximum segment lengths.
| Technology | Speed (Mbps) | Maximum Segment Length (meters) |
|
Token-Based |
||
| Token Ring | 4, 16 | 100 |
|
Simple Ethernet |
||
|
10base-T |
10 |
100 |
|
10base-F (multimode) |
10 |
up to 2,000 |
|
10base-F (single mode) |
10 |
up to 25,000 |
|
10base-5 |
10 |
500 |
|
10base-2 |
10 |
185 |
|
10base-36 |
10 |
3,600 |
|
Fast Ethernet |
||
|
100base-T4 |
100 |
100 |
|
100base-TX |
100 |
100 |
|
100base-FX (multimode) |
100 |
412 (1/2 duplex) |
|
2,000 (full duplex) |
||
|
100base-FX (single mode) |
100 |
20,000 |
|
100VG |
100 |
media |
|
Miscellaneous |
||
|
ATM |
155 to 622 |
media |
|
FDDI (single mode) |
100 |
40,000-60,000 |
|
FDDI (multimode) |
100 |
2,000 |
|
FDDI (TP) (CDDI) |
100 |
100 |
|
Fibre Channel |
133 to 1000 to 1250 |
10,000 |
|
Gigabit Ethernet |
||
|
1000base-T (UTP) |
1000 |
100 |
|
1000base-T (fiber--single mode) |
1000 |
3,000 |
|
1000base-T (fiber--multimode) |
1000 |
500 |
|
1000base-T (coax) |
1000 |
25 |
© Copyright, Macmillan Computer Publishing. All rights reserved.
){kind=link}
){kind=link}
){kind=link}
){kind=link}
){kind=link}
){kind=link}
){kind=link}
){kind=link}
){kind=link}
){kind=link}
){kind=link}
){kind=link}
){kind=link}
){kind=link}