Data communicationA communication model (elements of data communication)

The fundamental purpose of a communication system is the exchange of data between two parties. E.g. the exchange of data between server and a workstation over a public telephone line. Another example is the exchange of voice signals between two telephones over the same network. The key elements of data communication model are as follows:

Source: This device generates the data to be transmitted. Examples are of telephones and personal computers.

Transmitter: this device transmits the data generated by the source device. Usually, the data generated by a source system are not transmitted directly in the form in which the data was generated. Rather, a transmitter transforms and encodes he data in such a way that data can be transmitted across a transmission system (media). E.g. a modem takes digital data from a computer and transforms that digital bit stream into an analog signal that can be handled by the analog telephone network.

Transmission system: this is the path the data follows to reach the destination device. This can be a single transmission line or a complex network connecting source and destination devices.

Receiver: the receiver accepts the signals from the transmission system and converts it into a form that can be handled by the destination device. For example, a modem will accept analog signals from transmission line (telephone network) and will convert it into a digital bit stream so that a device such as a computer can handle it.

Destination: this device takes the incoming signals from the receiver.

Data communication networksIn its simplest form data communication takes place between two devices that are directly connected by some form of point-to-point transmission medium. Often, however, it is impractical for two devices to be directly, point-to-point connected because of the following reasons:

1. The devices are very far apart. It would be vary expensive, for example, to provide a dedicated link between two devices thousands of miles apart.

2. There is a set of devices, each of which may require a link to many other devices at various times. Examples are all of the telephones in the world and all of the terminals and computers owned by an organization. Except for the case of a very few devices, it is impractical to provide a dedicated link between each pair of devices.

The solution to this problem is to attach each device to a communication network.

Communication networks are traditionally classified into the following two major categories:

LAN (Local area networks): the scope of a local area network is small. A local area network is a number of computers and other devices connected to each other

1

1

by cable in a single location, usually a single floor of a building or all the computers in a small company.The internal data rates (speed) of a local area are much greater than those of the wide area networks. Therefore local area networks are more suitable for resource sharing between multiple computers.

Traditionally, local area networks make use of a broadcast approach rather than a switching approach unlike wide area networks. In a broadcast communication network, there are no switching devices. At each station (computer), there is a transceiver (transmitter/receiver) that communicates over a medium shared by other stations. A transmission from any one station is broadcast to and received by all other stations.

WAN (Wide-Area networks): Wide-Area networks have traditionally been considered to be those that cover a large geographical area. A WAN consists of a number of interconnected switching devices, which route data from a source device to a destination device. Stated simply wide area networks are the set of connecting links between different local area networks. These links are made over telephone lines leased from various telephone companies. Wide area networks can also be created with satellite links, packet radio or microwave transceivers but these options are generally far more expensive than leased telephone lines, but they can be used in areas where leased lines are not available.

The speed offered by wide area networks is much slower than the slowest local area networks. This makes the sharing of resources over a wide area network difficult. Generally, wide area networks are used for exchange of short messages such as e-mail or html traffic.

Advantages of a networkNetworks provide the following advantages:

1. Information sharingNetworks provide the facility of centrally controlling and sharing information. One or more computers in a network can be used to store the shared information and have all other computers on the network access that shared information. The computer or the computers on which the shared information is stored help centralize the information and maintain control over it. The computers, which store the shared information, are called servers and special software and operating systems are used in server computers.

2. Sharing hardware resourcesComputers that are not networked cannot effectively share hardware resources. For instance, a small office with ten stand-alone computers and one printer allows only the user with the printer attached to his or her computer to print. Other users must put their data on floppy disks, transfer it to the computer with the printer, and print it from there.

2

2

A network allows anyone connected to the network to use the printer. Not just the individual sitting at the computer to which the printer is attached. Network computers can also share the following devices:

Fax modems Scanners Hard disks Floppy disks CD-ROMs Tape backup units Plotters Almost any other device that can be attached to a computer

3. Sharing software resourcesSoftware resources can be used more effectively over a network. With stand-alone computers software must be present on each computer’s hard disk, whether or not that computer is used at that for that task or not. Software costs can increase for a large number of computers. It is also difficult and time consuming to install and configure the software individually on every one of the computers.With a network, software can be installed and configured centrally on a single computer, vastly reducing the cost and the work required to make computer programs available to an organization.

4. Preserving informationA network allows for information to be backed up to a central location. Important information can be lost by mistake or accident on a stand-alone computer that has no back up. It is difficult to maintain regular back ups on a number of stand-alone computers. When information is stored in a central location then all the important information can be backed up by backing up the central location (server) only.

5. Protecting informationA network provides a more secure environment for a company’s information. With stand-alone computers, access to computers often means access to the information on the computers. Networks provide an additional layer of security by the use of passwords. Each user can be assigned a different account name and password, allowing network servers to distinguish among those who need access to have it and protecting information from tampering from those who do not.

6. E-MAILThe computer network can also help people communicate. One of the greatest benefits to the users of networks is electronic mail or e-mail. E-mail is the fastest possible method of exchanging mails.

Roles of computers in a LANThere are three roles for computers in a LAN, which are as follows: Clients, which use but do not provide network resources. Peers, which both use and provide network resources. Servers, which provide network resources.

3

3

The type of operating system the computer uses determines each of these computer roles. Servers use network operating system such as Novell Netware or Windows NT server. Client computers use client operating system, such as MS-DOS or OS/2 2.0. Peers run peer network operating system such as Windows 95 or Macintosh operating system. Each of this operating system is optimized to provide service for the role it plays.

Based on the roles of computers attached to them, networks are divided into three types:

1. Server based or client-server networks contain clients and servers that support them.

2. Peer-to-peer or peer networks, which have no servers and use the network to share resources among independent peers.

3. Hybrid networks, which is a client-server network that also has peers sharing resources.

Server-based networks and domainsServer based networks are defined by the presence of servers on the network that provide security and administration of the network. Server based networks divide the processing tasks between clients and servers. Clients also referred to as front-end request for services, such as file storage and printing, and servers often called back-end deliver them. Servers are typically more powerful than client computers.In Windows NT server based networks are divided into domains. Domains are collections of networks and clients that share security trust information. The shared security trust is stored on special servers called domain controllers. Domain controllers control domain security and logon authentication. There is one master domain controller called primary domain controller (PDC), which may be assisted by secondary domain controllers called backup domain controller (BDC) during busy or when PDC is not available. None of the users in a domain can access the network resources in a domain until being authenticated by a domain controller.

4

4

Figure Figure 11 showing a client-server network with a shared printer showing a client-server network with a shared printer

Advantages of a server-based networkServer-based networks have the following advantages:

1. Strong central security.2. Central file storage, which allows all users to work from it set of data

and provides easy backup of critical data.3. Ability of servers to pool available hardware and software, decreasing

overall costs.4. Optimized dedicated servers, which are faster than peers at providing

resources.5. Less intrusive security, since a single password allows access to all

shared resources on the network.6. Freeing of users from the tasks of managing resource sharing.7. Easy manageability of a large number of users.8. Central organization, which keeps data from getting lost among

computers.

Disadvantages of a server-based networkAll the disadvantages of server-based networks are related with its high cost. Server-based networks are expensive because of the following reasons:

1. Expensive dedicated hardware is required.2. Expensive network operating system software and client licenses.3. A dedicated network administrator is also usually required.

Peer networksPeer networks are defined by the lack of central control over the network due to the absence of servers. Since there are no servers in a peer network, therefore users simply share disk space and resources such as printers and faxes as and when needed.

Peer networks are organized into workgroups. Workgroups have very little security control. There is no central login process. If a user has logged into one peer computer on the network, that user can then any resources on the network that are not controlled by a specific password.

Access to individual resources can be controlled, if the user who shared a resource makes the resource password protected. Because there is no central security trust, users will have to remember individual password, which the user wants to access. This can be quite inconvenient.

Peers are also not optimized to share resources. Generally, when a number of users are accessing resources on a peer, the user of that peer will notice significantly degraded performance. Peers also have licensing limitations that prevent more than a small number of users from simultaneously accessing resources.

5

5

Peer networks are recommended for use in small networks where the number of computers is less than 10 because administration in a peer network is decentralized due to the absence of server computers.

Figure Figure 22 showing a peer network showing a peer network

Advantages of peer networksPeer networks have many advantages, especially for small businesses that cannot afford to invest in expensive server hardware and software:

1. No extra investment in server hardware or software is required.2. Easy setup.3. No network administrator required.4. Ability of users to control resource sharing.5. No reliance on other computers for their operation.6. Lower costs for small networks.

Disadvantages of peer networksPeer networks have the following disadvantages:

1. Additional load on computers because of resource sharing.2. Inability of peers to handle as many network connections as servers.3. Lack of central organization, which makes data hard to find.4. No central point of storage for backing up data.5. Requirement that users administer their own computers.6. Weak and intrusive security.7. Lack of central management, which makes large peer networks hard

to work with.

Hybrid networksHybrid networks have all three types of computers operating on them and generally have active domains and workgroups. This means that while most shared resources are located

6

6

on servers, network users still have access to any resource being shared by a peer in a workgroup. It also means that network users do not have to logon to the domain controllers in order to access the resources of a peer in workgroup.

Advantages of hybrid networksHybrid networks have the following advantages:

1. The advantages of server-based networks.2. Many of the advantages of the peer networks.3. Ability of users and network administrators to control security based

on the importance of the shared resource.

Disadvantages of hybrid networksHybrid networks share the disadvantages of the server-based networks.

Network topologyThe way in which computer connections are made is called the topology of the network. The physical layout of cabling is called the topology or the physical shape of network is called its topology. It is important to select the right topology for the network. Each topology has its own strength and weaknesses. The four most common topologies are as follows:

i. Bus topologyii. Star topology

iii. Ring topologyiv. Mesh topology

Bus topologyThe bus topology is often used when the network installation is small, simple or temporary.

Figure 4 showing a bus networkFigure 4 showing a bus network

A single cable connects all the computers in a bus network. Bus is a passive topology in which computers only listen or send data. They do

7

7

not take data and send it on or regenerate it. So if one computer on the network fails, the network is still up. Since bus is a passive topology therefore it experiences signal loss.

When a computer sends a signal on a bus network, all the computers receive that signal, but only one (the one with the address that matches the one encoded in the message) accepts the information. The rest disregard or ignore the message.

Only one computer at a time can send a message, therefore the number of computers attached to a bus network can significantly affect the speed of the network. A computer must wait until the bus is free to transmit data. These factors also affect star and ring networks.

Another important issue in bus network is termination. If a signal gets to the end of a cable, it bounces back on that cable. When a signal echoes back and forth along an unterminated bus, it is called ringing. To prevent the signal from bouncing up and down the cable (ringing), devices called terminators have to be attached to both ends of the cable. A terminator absorbs an electronic signal and clears the cable this way so that other computers can send data on the network. Cables cannot be left unterminated in a bus network. Bus topology network does go down when the cable gets disconnected between one of the workstations as shown in the Figure.

Figure Figure 33: a cable break can bring down the entire bus network: a cable break can bring down the entire bus network

8

8

Advantages of the busBus network has the following advantages:

1. The bus is simple, reliable in very small networks, easy to use, and easy to understand.

2. The bus requires the least amount of cables to connect the computers together and is therefore the least expensive topology.

3. It is easy to extend a bus. Two cables can be joined into a one longer cable with a BNC barrel connector, making a longer cable and allowing more computers to join the network.

4. A repeater can also be used to extend a bus. A repeater regenerates the signal allowing it to cover longer distances.

Figure Figure 44: a repeater regenerates the signal: a repeater regenerates the signal

Disadvantages of the busBus topology has the following disadvantages:

1. Heavy network traffic can slow down a bus considerably, because any computer can transmit at any time and computers on most bus networks do not coordinate with one another to reserve times to transmit, a bus network with a lot of computers can spend a lot of its bandwidth (capacity for transmitting information) with the computers interrupting each other instead of communicating.

2. Each barrel connecter weakens the electrical signal and too many may prevent the signal from being correctly received all along the bus.

3. It is difficult to troubleshoot a bus. A cable break or a loose connecter will cause reflections and will bring down the whole bus network, causing all network activity to stop.

Star topology In a star topology, all the cables run from the computers to a central location, where they are all connected to a central device called a hub.

9

9

Stars are used in concentrated networks, where the endpoints are directly reachable from a central location, when network expansion is expected, and when the greater reliability of star network is needed.

Each computer on a star network communicates with a central hub that resends the message either to all the computers (in a broadcast network) or only to the destination computer (in a switched star network). The hub in a broadcast network can be active or passive whereas switched star networks have intelligent hubs or switches operating on them.

Passive hubThe function of a passive hub is simply to receive data from one port of the hub and send it out to the other ports. For example, an 8-port hub receives data from port 3 and then resends that data to ports 1, 2, 4, 5, 6, 7, and 8. It is as simple as that.

Active hub (Multi-Port Repeaters)

An active hub provides the same functionality of a passive hub with an additional feature. Active hubs repeat the data while resending it to all of the ports. By using active hubs you can increase the length of your network. It is important to remember that UTP (unshielded twisted pair) Category 5 cabling can be run a maximum of 100 meters. With an active hub, you can run Category 5 UTP 100 meters on each side of the hub.

Hybrid hub

A hybrid hub is a hub that can use many different types of cables in addition to UTP cabling. A hybrid hub is usually cabled using thinwire or thickwire Ethernet. Hybrid hubs are the most common type of hub. Hybrid hubs are used to interconnect hubs that are further than the 100-meter.

Advantages of star network

i. It is easy to modify and add new computers to a star network without disturbing the rest of the network. Computers can be attached to a star network simply by running a new line (cable) from the computer to the central hub.

ii. The center of a star network is a good place to diagnose network faults. Intelligent hubs (hubs with microprocessors) also provide for centralized monitoring and management of the network.

10

10

iii. Single computer failure does not bridge down the whole star network.

iv. Using the hybrid hub provides the facility of using several cable types in the same network.

Disadvantages of star network

1.If the central hub fails, the whole network fails to operate.

2.Many star networks require a device at the central location to rebroadcast or switch network traffic.

3.It costs more to cable a star network because every computer is connected to the hub with separate cables.

Star BusIf you replace the computers in a bus topology with the hubs from star topology networks, you get a star bus topology as illustrated in Figure.

11

11

Figure Figure 55: showing a star bus network: showing a star bus network

Star RingFigure displays a star ring, also called a star wired ring. The smaller hubs are internally wired like a ring and connected to the main hub in a star topology.

Figure Figure 66: showing a star ring network: showing a star ring network

Ring network

In a ring network each computer is connected to the next computer, with the last connected to the first.

12

12

Figure Figure 77: showing a ring network.: showing a ring network.

Rings are used in high performance networks, requiring that bandwidth be reserved for time sensitive features such as support for audio and video, or when even performance is needed for a large number of clients across the network.

In a ring topology, all computers are connected with a cable that loops around. Like a circle that has no start and no end, terminators are not necessary in a ring topology. Signals travel in one direction on a ring while they are passed from computer to the next as illustrated in Figure. Every computer checks the packet for its destination and passes it on as a repeater would. If one of the computers fails, the entire ring network goes down.

Some ring networks do token passing. A short message called a token is passed around he ring until a computer wishes to send data to another computer. That computer modifies the token, adds the destination device’s address and data and sends it around the ring. Each computer in sequence receives the token and passes the token to the next computer until either the address enclosed with the token matches the address of a computer or the token returns to its origin. The receiving computer sends a message to the source computer indicating that the message has been received. The source computer then creates another token and places it on the network, allowing other computers to capture the token and begin transmitting. The token circulates in the ring until a computer is ready to send data.

13

13

This all happens very quickly. A token can circle a ring 200 meters in diameter at about 10,000 times a second.

Advantages of the ring network1. Because every computer is given access to the token, no one

computer can monopolize the network.

2. The fair sharing of the network allows the network to degrade gracefully (continue to function in a useful, if slower manner rather than fail once capacity is exceeded) as more users are added.

Disadvantages of the ring network1. Failure of one computer on the ring can affect the whole network.

2. It is difficult to troubleshoot a ring network.

3. Adding or removing computers disrupts the ring network.

Mesh topologyThe mesh topology is distinguished by having redundant (repeated or more than one links) links between devices. A true mesh configuration has a link between each device in the network. This gets unmanageable beyond a very small number of devices. Therefore most mesh networks are not true mesh networks; rather they are partial mesh networks, which contain some redundant links but not all.

Mesh installationMesh topology networks become more difficult to install as the number of devices increase because of the sheer quantity of connections that must be made. A true mesh network of sic devices would require 15connections (5+4+3+2+1=15). A true mesh topology of seven devices would require 21 connections (6+5+4+3+2+1=21) and so on.

Mesh troubleshooting and reconfigurationMesh networks are easy to troubleshoot and are very fault tolerant. Media failure has less impact on mesh topology than on any other topology. The redundant links enable data to be sent over several different paths.Reconfiguration, like installation, gets progressively more difficult as the number of devices increase.

14

14

Advantages and disadvantages of mesh topologyThe major advantage of mesh topology is fault tolerance. Other advantages include guaranteed channel capacity and the facts that mesh networks are relatively easy to troubleshoot.Disadvantages include the difficulty of installation and reconfiguration, as well as the cost of maintaining redundant links.

Signal transmissionSignaling is the way data is transmitted across the transmission medium. Transmission medium may be guided (cable media) or unguided (wireless media). In both cases communication is in the form of electromagnetic waves. With guide media waves are guided along a physical path; examples of guided media are twisted pair, coaxial cable and fiber optics cable. Unguided media provide a means for transmitting electromagnetic waves but do not guide them; examples are propagation through air and vacuum.

Whenever data is to be transmitted along media, somehow the data, or the bits and bytes, must be represented in such a way that the sender can create a message and the destination device can understand it. This is done by means of encoding or modulation. The original signal is altered in a certain way o allow it to represent data.

The information or data to be communicated can exist in either of two forms:

Analog data Digital data

Analog dataAnalog data takes on continuous values on some interval. For example, voice and video are continuously changing patterns of intensity. Most data collected by sensors, such as temperature and pressure, are continuous-valued (digital). Another example of analog data is an analog clock. It is always changing its representation of time because the second hand never stops.

The most familiar example of analog data is audio or acoustic data, which, in the form of sound waves, can be perceived directly by human beings. Frequency components of speech may be found between 20Hz and 20 KHz.

Another common example of analog data is video. Here it is easier to characterize the data in terms of viewer (destination) of the TV screen rather than the original scene (source) recorded by the TV camera. To produce a picture on the screen, an electron beam scans across the surface of the screen from left to right and top to bottom. As the beam scans, the analog value changes. The video image, then, can be viewed as a time-varying analog signal.

Digital dataDigital data takes on discrete values. It represents either one value or the other not anything in between, for example, on or off, true or false, 1 or 0 and so on. Text and

15

15

integers are familiar examples of digital data. Similarly a digital clock does not shoe the variation of time between minutes. Its either 12:01 or 12:02, not anything in between.

A familiar example of digital data is text or character strings. While textual data are most convenient for human beings, they cannot, in character from, be easily stored or transmitted by data transmission and communication systems. Such systems are designed for binary data. Thus, a number of codes have been devised by which characters can be represented by a character of bits. The most commonly used code is the ASCII (American Standard Code for Information Interchange).

SignalsIn communication systems, data are propagates from one point to another by means of electric signals. The two signaling methods correspond to the two types of data (digital and analog):

Digital signalingAnalog signaling

Digital signalingA digital signal is a sequence of voltage pulses that may be transmitted over a wire media. Each pulse is a signal element. For example a constant positive voltage pulse may represent a binary 1, and a constant negative voltage pulse may represent a binary 0. Digital signals represent discrete states and the state change is practically instantaneous.

Because most computers are inherently digital, therefore most computer networks use digital signaling. There are many methods of encoding data in a digital signal. These methods are called encoding schemes. They can be grouped into two general categories, based on whether the recognition of a given state is triggered by certain voltage level or by the transition from one state to another:

Current state encoding schemesIn current state encoding strategies, data is encoded by the presence or absence of a signal characteristic or state. For example, a voltage of +5 might represent a binary 0, while a voltage of –5 might represent a binary 1.the signal is monitored periodically by network devices in order to determine its current state. That state then indicates the data. Unipolar is man example of current state encoding scheme.

UnipolarUnipolar is an encoding scheme that uses two levels for encoding data. One of the levels is zero, which could represent a binary 1, and the other level can either be positive or negative. If a particular implementation of Unipolar is using negative voltages, a –3V for example, would represent the other value, a binary 0.

Unipolar is not self-clocking, therefore a separate channel has to be used for providing the clocking.

State-Transition Encoding

16

16

State-transition encoding differ from current state methods in that it uses transitions in the signal to represent data, as opposed to encoding data by means of a certain voltage level or state. For example a transition occurring from high to low voltage could represent 1, while a transition from low to high voltage could represent a 0.

Another variation might be that the presence of a transition represents a 1 and the absence of a transition represents a 0. Manchester is an example of state-transition encoding.

Manchester In Manchester encoding, a low to high mid bit transition represents one value, such as a binary 0 and a high to low transition represents the other, such as the binary 1.Manchester encoding is used in Ethernet LANs. Due to the mid bit transitions Manchester is self- clocking.

Advantages of digital signalsIn general digital signals provide the following advantages over analog signals:

1. Fewer errors from noise and interference.2. Uses less expensive equipment.

On the other hand, one disadvantage is that digital signals suffer from greater attenuation than analog signals over the same distance.

Analog signalsAn analog signal consists of electromagnetic waves. An analog signal is a continuously varying electromagnetic wave that may be propagated (transmitted) over a variety of media. An analog signal is always changing and represents all values in a given range.

A wave cycle is the change from high to low and back to high (or low to high and back to low). Three characteristics are used to measure or describe electromagnetic waveforms: amplitude, frequency and phase.

Amplitude measures the strength of the signal or the height of the wave. Amplitude is measured in volts for electrical potential, amps for electric current, watts for electric power and decibels to indicate the ratio between powers of two signals.

Frequency is the amount of time it takes for a wave to complete one cycle. For example, if a signal takes 1 second to go from high to low and back to high (in other words complete one cycle), the frequency of the wave is one. Frequency is measured in hertz (Hz), or cycles per second.

Phase is a different type of measurement than amplitude or frequency in that it requires more than one wave. Phase is relative state of one wave, when timing began, relative to another reference wave. Phase is measured in degrees. The easiest phase shift to spot visually is that of 180 degrees.

Analog signal modulationAll the three characteristics of an analog wave i.e. amplitude, frequency and phase can be used to encode digital data in an analog signal. There are three main strategies for encoding digital data in analog signal. The first two amplitude shift key and frequency

17

17

shift key are considered current state encoding schemes and the third strategy is considered state-transition encoding.

Amplitude shift key (ASK)Amplitude shift key can be used to encode binary data by varying the amplitude of the signal. For example a stronger voltage could represent a binary 1 and a weaker voltage could represent a binary 0.

Frequency Shift Key (FSK)Frequency shift key, or FSK, is similar to ASK except that frequency not the amplitude of analog signal is changed to represent digital data. For example a higher frequency could represent a binary 1 and a lower frequency could represent a binary 0.

Phase shift key (PSK)Phase shift key uses a transition or shift from one phase to another to encode data. As in other state-transition encoding schemes, the presence or absence of a transition can be used to encode data. For example presence of transition can represent a 1 and the absence of transition can represent a binary 0.

Advantages of analog signalsIn general, analog signals provide the following advantages:

1. Less attenuation than digital signals over the same distance.2. Can be multiplexed to increase bandwidth.

One disadvantage is that analog signals are more prone to errors from noise and interference.

Bit synchronizationThere are different ways of encoding data in digital or analog signals. These encoding schemes rely on changes or modulations to a particular characteristic of the signal. The receiving network device must then interpret the signal by measuring that changed or modulated characteristic. Timing is important because the receiver needs to know, when to measure the signal in order to extract the correct meaning from it.

For two devices to communicate, a high degree of cooperation is required. Typically, data is transmitted one bit at a time over the transmission medium. The timing (rate, duration, spacing) of these bits must be the same for transmitter and receiver. The coordination of signal measurement timing is called bit synchronization. The two major methods of providing bit synchronization are asynchronous and synchronous.

Asynchronous bit synchronizationThe strategy with this scheme is to avoid the timing problem by not sending long, uninterrupted streams of bits. Instead, data is transmitted one character at a time, where each character is five to eight bits in length. Asynchronous communication requires that messages begin with a start bit with a value of binary 0. So that the receiving device can synchronize its internal clock with the timing of the message. When no data is being transmitted, the media is idle (the definition for idle is equivalent to the signaling element of binary 1) and the sender and receiver’s clocks are not synchronized.

Five to eight bits that actually make up the character follow the start bit. Usually, this is followed by a parity bit, which provides error checking. The parity bit is set by the

18

18

transmitter such that the total number of ones in the character, including the parity bit, is even (even parity) or odd (odd parity). A stop bit indicates the end of the transmission, which is a binary 1.

Synchronous bit synchronizationWith synchronous transmission, a block of bits is transmitted in a steady stream without start and stop bits. The block may be many bits in length. To prevent timing drift between sender and receiver, their clocks must somehow be synchronized. The following three methods are used for synchronous timing coordination.

1. Guaranteed state change2. Separate clock signals3. Oversampling

Guaranteed state changeGuaranteed state change describes a method in which the clocking information is embedded in the data signal. This way, the receiver is guaranteed that transitions will occur in the signal at predefined intervals. These transitions allow the receiver to continually adjust its internal clock.

The guaranteed state change is the most common method, and it is frequently used with digital signals. All the digital encoding schemes, which are self-clocking, use this method. For example, Manchester encoding uses guaranteed state change.

Separate clock signalsIn the separate clock signals method, a separate channel between the sender and receiver provides the clocking information. Since this method requires twice the channel capacity of embedding the clock in the data stream, it is insufficient. This method is most efficient for shorter transmissions such as those between a computer and a printer.

OversamplingOversampling is a method in which the receiver samples the signal at a much faster rate than the data rate. This permits the use of an encoding method that does not add clocking transitions. If the receiver samples the signal ten times more quickly than the data rate, out of any ten measurements, one would provide the data information, and the other nine would determine whether the receiver’s clock is synchronized.

Signal TransmissionTwo techniques are used to transmit the encoded signals over cable: baseband and broadband. Baseband transmission uses digital signaling and broadband transmission uses analog signaling.

19

19

BroadbandBroadband LANs are the exception rather than the rule, although many cable companies are now offering Internet services at extraordinary transmission speeds. These networks differ from their baseband counterparts in that they use coaxial or fiber-optic cable to carry multiple channels of data. A single cable can carry five or six separate communications channels. A broadband network is analogous to cable TV: One cable brings in many TV channels as well as networking services.

Broadband systems use analog signaling and a range of frequencies. With analog transmission, the signals employed are continuous and nondiscrete. Signals flow across the physical medium in the form of electromagnetic or optical waves. Broadband technologies use amplifiers to bring analog signals back up to their original strength.

BasebandMost LANs are baseband. A baseband network uses only one channel on the cable to support digital transmission. Signals flow in the form of discrete pulses of electricity or light. With baseband transmission, the entire capacity of the communication channel is used to transmit a single data signal.

As the signal travels along the network cable, it gradually decreases in strength and can become distorted. If the cable length is too long, resulting in a signal that is weak or distorted, the received signal may be unrecognizable or misinterpreted. As a safeguard, baseband systems sometimes use repeaters to receive an incoming signal and retransmit it at its original strength and definition to increase the practical length of a cable segment.

Network media typesComputers send electronic signals to each other using electric current, radio waves, microwaves, or light-spectrum energy. These signals represent network data as binary pulses. The physical path through which computers send and receive these signals is called the transmission media.

Transmission media are divided into two categories:

Cable mediaCable media have a central conductor enclosed in a plastic jacket. They are typically used for small LANs. There are three primary types of physical media that can be used at the Physical Layer: coaxial cable, twisted-pair cable, and fiber-optic cable. Transmission rates that can be supported on each of these physical media are measured in millions of bits per second (Mbps).

20

20

Wireless mediaWireless media typically use higher electromagnetic frequencies, such as radio waves, microwave, and infrared. Wireless media are necessary for networks with mobile computers or networks that transmit signals over large distances.

Characteristics of mediaEach media has certain characteristics that make it suitable for particular networks. To choose the best type of media, it should be known that how each medium characteristics relate to the following factors.1. CostThe cost of each media type should be weighed against the performance it provides, and the available resources. For example, it is a common practice among network integrators to attempt to run a network across unused, left over telephone cabling. Although this reduces costs, in many cases it can prove to be a wrong solution. For example, when cable drops of greater than100 meters are required. Each network installation is different and should have the most affordable viable solution. When deciding upon a network media type, real needs of the network should be considered. For example fiber optics is fast, but a network may not need that much of speed.2. InstallationThe difficulty in installation depends on the individual situation, but some general comparisons between the media are possible. Some types of media can be installed with little tools and less training; other requires more training and knowledge and may be better left to professionals. For example, unshielded twisted pair cable is easy to install, but fiber optics cable requires professional training. 4. Bandwidth capacityThe capacity of a medium is usually measured in bandwidth. In networking terms, bandwidth is measured in mega bits per second (mbps). A medium with a high capacity has a high bandwidth, and a medium with low capacity has a low bandwidth. A high bandwidth normally improves the throughput and performance of a certain media type.5. Node capacityNode capacity is the maximum number of nodes that can be attached to a certain network media before expensive devices such as bridges, routers, repeaters and hubs must be used to expand the network.6. AttenuationElectromagnetic signals tend to become weak during transmission. This is referred to as attenuation. As the signals pass through the transmission medium, part of their strength is absorbed or misdirected.

21

21

This phenomenon imposes a limit on the distance a signal can travel through a medium without unacceptable degradation. When the signal gets weak, it becomes difficult to distinguish between a 1 and a 0 due to which errors in communication can take place. Because of attenuation and dispersion, it must be made sure that networks cables do not exceed the maximum length recommended for that cable.7.Electromagnatic interferenceElectromagnetic interference (EMI) effects the signal that is sent through the transmission media. EMI is caused by the outside electromagnetic wave effecting the signal, making it more difficult for the receiving computer to decode the signal. Some media are more influenced by EMI than others. EMI is often referred to as noise. A related concern is eavesdropping, especially if your network data requires a high level of security. The same characteristic of the cable that allow the external signal to interfere with the signal in the cable also make it easy for someone to detect the signal externally, without piercing the cable.

CoaxialCoaxial (or coax) cable looks like the cable used to bring the cable TV signal to your television. One strand (a solid-core wire) runs down the middle of the cable. Around that strand is insulation. Covering that insulation is a wire mesh and metal foil, which shields against electromagnetic interference, as illustrated in Figure. A final layer of insulation covers the wire mesh providing protection and insulation. Coaxial cable is resistant to the interference and signal weakening that other cabling, such as unshielded twisted-pair (UTP) cable, can experience. In general, coax is better than UTP cable in connecting longer distances and for reliably supporting higher data rates with less sophisticated equipment.

Just because the TV cable is coax does not mean it will work with computer networks. Coaxial cable comes in different sizes. It is classified by size (RG) and by the cable’s resistance to direct or alternating current measured in ohms called impedance, and the attenuation.

The following are some common coaxial cables commonly used in networks:

50-ohm, RG-8 and RG-11, used for thick Ethernet

50-ohm, RG-59, used for thin Ethernet

75-ohm, RG-59 used for cable TV

22

22

93-ohm, RG-62, used for ARC net

Figure 1: Coax cable

Thinnet Coaxial Cable

Thinnet refers to RG-58 cabling, is a flexible coaxial cable about ¼ inch thick. Thinnet is used for short distance communication and is flexible enough to facilitate routing between workstations. Thinnet connects directly to a workstation’s network adapter card using a BNC T-connector (See Figure 8-2) and uses the network adapter card’s internal transceiver. 10Base2 refers to Ethernet LANs that use Thinnet cabling.

Figure 2: BNC T-connector

Figure 8-3 illustrates a bus type of network called a local bus. At each end of the bus there is a terminating resistor, or a terminator, of 50 ohms. Each workstation’s network adapter is connected to the bus via a single cable, called a drop, using a BNC T-connector. The network adapter’s internal transceiver is a device that transmits and receives signals. Even if your network consisted of only two computers, T-connectors and terminates are still required.

23

23

 

Figure 3: Bus network

Thicket Coaxial Cable

Thicket coaxial cable can support data transfer over longer distances better than Thinnet and is usually used as a backbone to connect several smaller thinnet-based networks. The diameter of a thicket cable is about ½ inch and is harder to work with than a thinnet cable. A transceiver is often connected directly to thicknet cable using a connector known as a piercing tap. Connection from the transceiver to the network adapter card is made using a drop cable to connect to the adapter unit interface (AUI) port connector. 10Base5 refers to Ethernet LANs that use thicknet cabling. The following Figure illustrates an AUI connector.

Figure 9: AUI connector

Coaxial cable has the following characteristics. Cost

Coax is relatively inexpensive. The cost for thin coaxial cable is less than STP or category 5 UTP. Thick coaxial is more expensive than STP or category 5 UTP but less than fiber-optics cable.

InstallationInstallation is relatively simple. With a little practice, installing the connectors becomes easy, and the cable is resistant to damage.

Bandwidth capacityA typical data rate for today’s coaxial networks is 10 Mbps, although the potential is higher. Coaxial cable’s bandwidth potential increases as the diameter of the inner conductor increases.

Node capacityThe specified maximum number of nodes on a thinnet segment is 30 nodes and on a thicknet segment is 100 nodes.

Attenuation

24

24

Because it uses copper wire, coaxial cable suffers from attenuation, but much less so than twisted-pair cable. Coaxial cable runs are limited to a couple of thousand meters.

EMICoaxial cable is still copper wire and vulnerable to EMI and eavesdropping. However, the shielding provides a much better resistance to EMI’s effects.

Twisted-pair cableTwisted pair cable uses one or more pairs of two twisted copper wires to transmit signals. It is commonly use as telecommunication cable.When copper wires that are close together conduct electric signals, there is a tendency for each wire to produce interference in the other. One wire interfering with another in this way is called cross talk. To decrease the amount of cross talk and outside interference, the wires are twisted. Twisting the wires allows the emitted signals from one wire to cancel out the emitted signals from the other and protects them from outside noise.Twisted pairs are two color-coded, insulated copper wires that are twisted around each other. A twisted pair consists of one or more twisted pairs in a common jacket.There are two types of twisted pair cables: shielded and unshielded.

Unshielded twisted-pair cableUnshielded twisted-pair cable consists of a number of twisted pairs with a simple plastic casing. UTP is commonly used in telephone systems. Unshielded twisted-pair (UTP) cables are familiar to you if you have worked with telephone cable. The electrical industries association (EIA) divided UTP into different categories by quality grade. The ratings for each category refer to conductor size, electrical characteristics, and twists per foot. The following categories are defined:

Category 1 and 2 were originally meant for voice communications and can support only low data rates, less than 4 Mbps. These cannot be used for high-speed data communication. Older telephone networks used category 1 cable.

Category 3 is suitable for most computer networks. Some innovations can allow data rates much higher, but generally category 3 offers data rates up to 16 Mbps. This category of cable is the kind currently used in most telephone installations.

Category 4 offers data rates up to 20 Mbps. Category 5 offers enhancements over category 3, such as

support for fast Ethernet, more insulation, more twists per foot

25

25

and data rates of 100 Mbps and higher, but category 5 requires compatible equipment and more difficult installation. In a category 5 installation, all media, connectors, and connecting equipment must support category 5, or performance will be affected.

Data-grade UTP cable (category 3,4, and 5) consist of either 4 or 8 wires. Network topologies that use UTP require at least two pair wire.Because UTP cable was originally used in telephone systems, UTP installations are similar to telephone installations. For a four-pair, a modular RJ-45 telephone connector is used. For a two-pair cable a modular RJ-11 telephone connector is used.

Figure 10: Unjacketed twisted-pair cable and an RJ-45 connector

UTP’s popularity is partly because UTP was first used in telephone systems. In many cases a network can be run over the already existing wires installed for the phone system, at a great savings in installation.UTP cable has the following characteristics:

CostExcept for professionally installed category 5, UTP cabling is the least expensive medium but requires an additional component, a hub.

InstallationUTP cable is easier to install than coaxial because you can pull it around corners more easily. UTP cable’s installation can be done with very little training.

Bandwidth capacityThe most common data rate is 10Mbps.

Node capacitySince only two devices can be connected together by a UTP cable, the cable does not limit the number of computers in a UTP network. Rather, the hub or hubs that connect the cables together limit it.

AttenuationUTP is normally restricted to distances of 100 meters.

EMITwisted-pair cable is more susceptible to interference and should not be used in environments containing large electrical or electronic

26

26

devices. In addition, because copper wires emit signals, UTP is susceptible to eavesdropping.

Shielded twisted-pair cableShielded twisted-pair cable differs from UTP in that it uses a much higher quality protective jacket for greater insulation. Thus, it is less subject to electrical interference and supports higher transmission speeds over longer distances than UTP. STP cable has a shield usually aluminum/polyester between the outer jacket and the wires. The shield makes STP less vulnerable to EMI.

STP has the following characteristics:

Cost

STP costs more than UTP and thin coaxial cable but less than thick coaxial and fiber optics cable.

Installation

The requirement for special connectors can make STP more difficult to install than UTP. Because STP is rigid and thick (up to 1.5 inches in diameter), it can be difficult to work with.

Node capacitySince only two devices can be connected together by an STP cable, the cable does not limit the number of computers in an STP network. Rather, the hub or hubs that connect the cables together limit it.

Attenuation

STP does not outperform UTP much in terms of attenuation. The most common limit is 100 meters.

EMI

The biggest difference between STP and UTP is the reduction of EMI> the shielding blocks a considerable amount of interference. However, since it is still copper wire, STP still suffers from EMI and is vulnerable to eavesdropping.

Fiber optic-cable

Optical fibers carry digital data signals in the form of modulated pulses of light. It is enormously more efficient than the other network transmission media.

27

27

An optical fiber consists of an extremely thin cylinder of glass or plastic, called the inner core that conducts light. A concentric layer of glass, known as the cladding, surrounds the Inner core that reflects the light back into the inner core. A plastic sheath surrounds each fiber. The sheath can be either tight or loose. Tight configuration completely surrounds the fiber with a plastic sheath and sometimes includes wires to strengthen the cable. Loose configuration has a space between the sheath and the outer jacket, which is filled with a gel or other material. There are two fibers per cable—one to transmit and one to receive. Optical fibers are smaller and lighter than copper wire. One optical fiber is approximately the same diameter as a human hair.

Figure 12: Fiber-optic cable

Optical fibers may be single-mode or multi-mode. Single-mode fibers allow a single light path and are typically used with laser signaling. Multi-mode fibers allow multiple light paths. Single-mode fiber can allow greater bandwidth and cable runs than multi-mode but is more expensive.

A typical LAN installation at a computer or a network device that has a fiber-optic network interface card (NIC). This NIC has an incoming interface and an outgoing interface. The interfaces are directly connected with fiber-optics cable with special fiber-optics connectors. The opposite ends of the fiber optics cable are attached to a connectivity device or a splice center.

Optical interface devices convert computer signals into light for transmission through the fiber. Conversely, when light pulses come through the fiber, the optical interfaces convert them into computer signals. For single-mode fibers, light pulses are created by injection laser diode (ILD), which create a higher quality of light. For multi-mode fibers, light emitting diodes (LEDs) are used.

Fiber-optics cable has the following characteristics:

28

28

Cost

Fiber-optics cable is slightly more expensive than copper cable, but costs are falling. Associated equipment costs can be much higher than the copper cable, making fiber-optics networks much more expensive. Single-mode fiber devices are more expensive and more difficult to install than multi-mode devices.

InstallationFiber-optics cable is more difficult to install than copper cable. Each fiber connection and splice must be carefully made to avoid obstructing the light path. Also, the cables have a maximum bend radius, which makes cabling much more difficult.

Bandwidth capacityBecause it uses light, which has a much higher frequency than electricity, fiber-optics cabling can provide much higher bandwidths. Current technologies allow data rates from 100 Mbps to 2Gbps.the data rate depends on the fiber composition, the mode, and the frequency. A common multi-mode installation can support 100 Mbps of data rates over several kilometers.

Node capacity

Since only two devices can be connected together by a fiber-optics cable, the cable does not limit the number of computers in a fiber-optics network. Rather, the hub or hubs that connect the cables together limit it.

AttenuationFiber-optics cable has much lower attenuation than copper wires, mainly because light is not radiated out in the way electricity is radiated from copper cables. Fiber-optics cable can carry signals over distances measured in kilometers.

EMIFiber-optics cable is not subject to electrical interference. In addition, it does not leak signals, so it is also immune to eavesdropping. This type of cable is ideal for high-voltage areas or in installations where eavesdropping could be a problem.

Network ModelsConnecting two or more computers together to exchange information is a relatively simple concept, but there is a lot that must happen behind the scenes to make it a reality. Designing networking into an operating system or developing a network standard is quite an undertaking.

29

29

All the network hardware designers, software programmers, and operating system architects don’t want to reinvent the wheel each time they develop a new piece of hardware, have an idea for a new protocol, or want to create the latest greatest operating system. To facilitate the design and operation of networking components, network models were created to provide a framework. All popular, open network standards are based on the models discussed in this chapter. Understanding these models and how they relate to various network standards will give you a better understanding of how networks work.

Network Models

Network models provide a standard framework to use when designing complex communication systems. Since all networks carry out many of the same functions, industry players have devised network models to simplify their design.

Models outline standard issues associated with network design and allow the designer to solve each issue separately, modularizing the solution. Rather than developing a solution from top to bottom, from the operating systems to the interface cards and wires on the network, a network model allows the designers to relegate different parts of the design to different people. It also allows them to use a proven design, rather than developing their own.

Keep in mind that a network model is a framework to use, not a concrete method. It is up to the implementers to decide which parts of the model are relevant to accomplish their goals.

Network Communication Basics

Network communication has one very simple goal: to send 1s and 0s from one computer to another, quickly and without error. While this may seem easy enough from the surface, look under the covers and you’ll see it is a complex process. Network communications have to take into account many variables to work reliably.

Representing and Transmitting Data

When actually sending information across a network, there are many ways to send the data. Computer data is nothing but 1s and 0s, and it is always sent one bit at a time. But different computers look at those 1s and 0s in completely different ways. It is up to the network standard to define the correct way to encode data, transmit it, and then decode it at the other end. There are some basic rules for interpreting those bits and bytes.

Computers can start at either end of a binary number when transmitting it across a network. This is known as bit order. When a computer starts with the first digit of a binary number, it is using the

30

30

most significant digit. If it starts at the last digit, it is using the least significant digit.

Larger binary numbers consist of two or more bytes. Just as a computer can start at either end of a single binary number, it can also start at either end of a group of bytes. This is known as byte order. IBM-compatible PCs contain little endian processors, which expect the last, or least significant byte of data to appear first. Apple Macintoshes, on the other hand, contain big endian processors, which expect the first, or most significant, byte to appear first.

When transmitting data on a network, there may be many different kinds of computers listening. It is necessary to define whether the first or last bit is most significant. Otherwise, different computers could interpret the same data in very different ways.

End-to-End and Point-to-Point Transmission

Some networks are only concerned with end-to-end communication, meaning the two ends of the conversation deal with each other directly. This is best when timing is of great importance (take your telephone, for instance).

But maintaining connections between computers is not an efficient use of network resources. It would be like having the phone off the hook all day long, when you actually want to talk for only a few minutes a day.

For this reason, many computer networks use point-to-point transmission methods, where there may be one to dozens of points between the two ends (email is a good example of this). Each point is only concerned with transferring data from itself to the next point downstream. After that, it is up to the next point to ensure that the data continues on its way to the final destination point. After data passes through a point, that point is available for other network chores, even though the data may not have reached its destination yet.

Fragmenting Data, Sequencing, and Reliability

Seldom is the data you want to send across the network a single manageable unit. In fact, as computers and networks get faster and have larger capacities, there is more data to send! Since most data can’t be handled in its entirety for transmission, most networks chop or fragment large pieces of data into more manageable units.

When data is fragmented, it is important to ensure that all the pieces make it to the other end in the right sequence. If they are not in order, it is sometimes possible to resequence the data into the right order. If the data can’t be put back into the right order, then it must be resent. Steps must be taken before transmission to label data fragments

31

31

before sending them so the receiving end can figure out what order they belong in.

Error Checking

We all know that we don’t live in a perfect world, and imperfection carries straight down into computer networks. Imperfections in network transmissions can result in corrupted, useless data on the receiving end. When sending data across a network, error checking can be used to ensure that the data received is identical to the data that was sent originally. Error checking can happen at many different levels of the communication process.

A basic method of checking for errors with transmitted data is the use of a parity bit. Before sending data, the numbers of individual bits that make up the data are counted. If there are an even number of bits, a parity bit is set to one and added to the end of the data so that the total of the bits being sent is odd. If there are an odd number of bits, the parity bit is set to zero and added to the end. The receiving computer adds up the bits received and if there are an even number of bits, the computer assumes that an error has occurred. The parity method is not foolproof, since if an even number of bits is corrupted, they will offset each other in the total.

The checksum is a form of error checking that simply counts the number of bits sent and sends this count along. On the receiving end, the bits are once again counted and compared with the original count. If they match, it is assumed the data was received correctly.

Another type of error checking is the cyclical redundancy check (CRC). This involves running a byte or group of bytes through a mathematical algorithm to produce a single bit or byte to represent the data (a CRC). The CRC value is transmitted with the data. When the data reaches its destination, the receiver runs it through the same mathematical algorithm. The results are compared with the original CRC, and if they match, the receiving computer assumes that data is correct. If they do not match, the receiver must discard the data and try again.

Sometimes the integrity of the data is checked at each step along the way (connection-oriented). At other times, there is no error checking on the network (connectionless); instead, the error checking is left up to the software sending the data.

OSI Model

The most common network model used in PC networks is the Open Systems Interconnect (OSI) model. The OSI model was developed from the late 1970s to its current form in the mid-1980s by the International Standards Organization (ISO). Both Microsoft, Novell, and all the major PC networking giants use the OSI model as a basis for network design.

32

32

The OSI model consists of seven layers. They cover all aspects of networking, from the topmost issues (“How do I print to the network printer?”) all the way down to the lowest technical issue (“What voltage at what frequency do I apply to which wire?”). The seven layers help break down the aspects into manageable units that interact with one another. The layers are, from top to bottom:

Application Presentation Session Transport Network Data Link Processing

A common mnemonic to help you remember the layers from top to bottom is, “All People Seem To Need Data Processing.” There are many more phrases, and you can always invent your own.

All the layers in the OSI model work in a hierarchy. If a computer is sending data, each layer receives the data from the layer above it, performs any applicable work on that data, and adds on its own information regarding that data. The layer then sends the data on down to the next layer.

When a computer is sending data, each layer receives the data from the layer beneath it, processes it, and sends it to the layer beneath it. The opposite occurs on the receiving computer. In Figure 2-1, the Application Layer on Computer 1 communicates with the Application Layer on Computer 2, going down the hierarchy, and then back up.

Figure 1: Communicating through the hierarchy of OSI layers

33

33

As data is sent down the hierarchy, each layer appends its own information to the data for processing by the same layer on the destination computer. Each layer adds a header as the data travels down through the OSI layers, and the associated layer on the receiver removes the headers.

Figure 2-2 demonstrates how each layer adds information to the data as it is sent, and conversely how that information is removed as the data is received.

Figure 2: Headers added removed by OSI layers

The seven layers can be broken into two groups to help further understand their basic functions. The first three layers (Application, Presentation, and Session) are primarily used by applications. The four lower layers (Transport, Network, Data Link, and Physical) are concerned with data transport, or simply getting data from one network device to another.

Functions of the Layers

Each layer has specific functions that it defines. Some functions are defined in more than one layer (such as error control and flow control). While this seems redundant, it does not mean that these functions must be implemented at both layers, no matter what. Don’t forget the OSI is a model. One designer may use error control at one layer; another may use it at a different layer. It all depends on the designer’s goals.

While it is very easy to memorize the various layers of the OSI model from top to bottom, it is a bit easier to learn about what these layers do by taking a bottom-to-top approach.

PhysicalThe bottom layer of the OSI hierarchy is only concerned with moving bits of data onto and off of the network medium. The Physical Layer

34

34

does not define what that medium is, but it must define how to access it. This includes the physical topology (or structure) of the network, the electrical and physical aspects of that medium used, and encoding and timing of bit transmission and reception.

Data LinkThe Data Link Layer handles many issues for communicating on a simple network (The Network Layer discussed in the next section performs the functions necessary to communicate beyond a single physical network.) This layer takes the frames generated by the upper layers and takes them apart for transmission. When receiving messages from the network, it reassembles this information back into frames to send to the upper layers. This layer actually does a lot more than just break apart and put together frames.

The 802 model breaks the Data Link Layer into two sub layers: logical link control (LLC) and media access control (MAC). The LLC layer starts and maintains connections between devices. When sending data from your workstation to a server on the same network segment, it is the LLC sublayer that establishes a connection with that server. The MAC layer allows multiple devices to share the media. Most LANs have more than one computer (of course!), and the MAC sublayer determines who may speak and when.

Another important job of the Data Link Layer is addressing. The MAC sublayer maintains physical device addresses for communicating with other devices (commonly referred to as MAC addresses). Each device on the network must have a unique MAC address, otherwise the network will not know exactly where to send information when a node requests it. For example, how would the postal service know where to send your bills without your address?

Most network interface cards (NICs) in a computer provide the MAC address as an address burned into the interface card. Some older network cards even required an administrator to set the address manually using switches. Even with a permanent MAC address burned into the card, some protocols allow you to define this address via software, although this is unusual.

The MAC address is used to communicate only on the local network. When transmitting to a server on the same LAN segment, the protocol uses the MAC addresses to communicate between the two computers. If the server is located on another network segment across a WAN, the MAC address of the nearest router (routers are discussed later in this chapter) is used to send the information, and it is up to the router to send the data further on.

35

35

Finally, the Data Link Layer manages flow control and error correction between devices in a simple network. In more complex internetworks, it is up to the Network Layer and other upper layers to perform these functions.

NetworkThe Network Layer is one of the most complex and important ones. The Network Layer manages addressing and delivering packets on a complex internetwork. Devices known as routers, which utilize routing tables and routing algorithms to determine how to send data from one network to another, join internetworks.

The most obvious example of an internetwork is the Internet. The Internet is very large, covering the entire globe, and consists of an almost every conceivable type of computer, from palmtops to mainframes.

In order to operate on an internetwork, each network that participates must be assigned a network address. This address differentiates each network from every other network that forms the internetwork. When sending data from one network to another, the routers along the way use the network addresses to determine the next step in the journey.

The Network Layer also allows the option of specifying a service address on the destination computer. All modern operating systems (UNIX, Windows NT, OS/2, etc.) run many programs at once. The service address allows the sender to specify which program on the destination the data being sent is for. Service addresses that are well defined (by networking standards, for example) are called well-known addresses. Service addresses are also called sockets or ports by various protocols.

TransportThe Transport Layer works hard to ensure reliable delivery of data to its destinations. The Transport Layer also helps the upper layers (Application, Presentation, and Session) communicate with one another across the network while hiding the complexities of the network.

The Transport Layer also interacts with the Network Layer, taking on some of the responsibilities for connection services.

One of the functions of the Transport Layer is segment sequencing. Sequence switching is a connection-oriented service that takes segments that are received out of order and resequences them in the right order.

36

36

Another function of the Transport Layer is error control. It commonly uses acknowledgements to manage the flow of data between devices. Some Transport Layer protocols can also request retransmission of recent segments to overcome errors.

SessionThe Session Layer manages dialogs between computers. It does this by establishing, managing, and terminating communications between the two computers. There are three types of dialogs that the Session Layer uses.

Simplex dialogs allow data to flow in only one direction. Since the dialog is one way, information can be sent, but not responded too, or even acknowledged. An example of a simplex dialog is a Public Announcement (PA) system in a large building. Announcements can be made, but the PA system doesn’t allow any response or acknowledgement.

Half-duplex dialogs allow data to flow in two directions, but only one direction at a time. With half-duplex dialogs, replies and acknowledgements are possible. But this isn’t always the most efficient method. If an error is detected early on in transmission, the receiver must wait for the sender to finish before any action can be taken. A CB radio is an excellent example of a half-duplex dialog.

Full-duplex dialogs let data flow in both directions simultaneously. This method allows more flexibility, but also requires more complex communication methods. A telephone is a prime example of full-duplex communication.

When a session is established, there are three distinct phases involved. Establishment is when the requestor initiates the service and the rules for communication are established. Once the rules are established, the data transfer phase may begin. Both sides know how to talk to each other, what the most efficient methods are, and how to detect errors, all because of the rules defined in the first phase. Finally, termination is when the session is complete and communication ends in an orderly fashion.

PresentationIt is up to the Presentation Layer to make sure that data sent by the Application Layer and received by the Session Layer is in a standard format. As discussed earlier, different types of computers can interpret identical data differently.

A network standard defines the proper format for any data as it is transmitted. When the Presentation Layer receives data from the

37

37

Application Layer to be sent over the network, it makes sure the data is in the proper format and if not, it converts the data. On the flip side, when the Presentation Layer receives data from the Session Layer from the network, it makes sure the data is in the proper format, and once again converts it if not.

ApplicationThe Application Layer provides a consistent, neutral interface to the network. Many people confuse the Application Layer with an actual software package, such as a word processor. This is not the case. The Application Layer provides a consistent way for an application to save files to the network file server or print to a network printer.

An example of this is how Windows 95 makes it just as easy to print to a network printer, as it is to print to a locally attached printer. This is the Application Layer in practice.

The Application Layer also advertises a computer’s available resources to the rest of the network.

Relationships Between Protocols and the OSI Layers

The various layers of the OSI model provide a basic framework for implementing real protocol stacks. You may notice some redundancies, however. This is not by accident. It is up to the designers to decide where certain functions make the most sense to deal with. Never forget that the model is a framework, not a concrete method of implementation!

The OSI Model and Addressing

In any network consisting of more than two computers, there needs to be a way to identify individual computers on the network. Unless you always want every computer on the network to use any data you are sending, you must be able to single out the computer you are trying to communicate with. This requires network addressing. The two layers of the OSI model that largely deal with addressing are the Data Link and Network Layers.

The Data Link Layer is only concerned with addressing on the local network. The MAC sublayer defines physical device addresses (or MAC addresses), which are used to uniquely identify computers on a network. These unique addresses can be used to identify the individual computers.

The Network Layer deals with addresses on a larger scale, handling internetworking between multiple networks. The Data Link Layer handles addressing individual computers, whereas the Network Layer handles addressing for individual networks. Once the destination

38

38

network is reached, the addressing at the Data Link Layer once again comes into play in order to find the exact computer the data is headed for.

Devices that Communicate at Each OSI Layer

There are many devices used on a network beyond our computer. Repeaters, bridges, and routers are devices used to link individual LANs together to form larger internetworks. When comparing the layers of the OSI model to the roles each of these devices, note that each one operates within a specific layer of the OSI model.

Physical Layer Devices

The Physical Layer is only concerned with moving the bits and bytes of data onto the network. This layer does not deal with any addressing issues. It gets data onto and off of the wire. Repeaters operate at the Physical Layer of the OSI model. They simply listen to all network traffic on one port and send it back out through one or more ports, extending smaller networks into a larger, single network, as illustrated in Figure 2-3.

Figure 3: Repeater linking two physically separate networks

Functions

A repeater simply receives frames, regenerates them, and passes them along. It performs no processing of the frames or the data

they contain. Because it is not performing much processing, repeaters are simpler in design, and therefore less expensive than other devices used to connect networks, like bridges and routers.

Figure 2-3 shows a repeater linking Network 1 to Network 2. Any data sent out on Network 1 is picked up by the repeater and sent out over Network 2 with no changes. The opposite happens when transmitting on Network 2. As far as all the devices on each network are concerned, there is only one network.

Relation to OSI Layer Functions

Because repeaters operate at the Physical Layer, they do not need any addressing data from the frame. Repeaters do not even look at the frames they are forwarding, passing along even damaged frames. This

39

39

can be especially problematic if one segment malfunctions and begins a broadcast storm. The repeater forwards all those erroneous broadcasts faithfully!

Examples

Repeaters are primarily used to extend a LAN beyond physical limitations. In a manufacturing setting, a computer on the plant floor may be further from the rest of the network than the physical limits of the media allow. Inserting a repeater between the computer and the rest of the LAN could allow the computer access to the network.

Repeaters can also join networks that use the same frame type but different types of cabling. Suppose the Marketing and Accounting departments each have their own LAN, both using the 802.3 frame type. Marketing uses twisted-pair Ethernet cabling and accounting uses thin coaxial cabling. If the two departments want a simple, inexpensive way to join their LANs, they can use a repeater, despite the different cabling. If Marketing uses a Token Ring network scheme and Accounting uses Ethernet, they cannot use a repeater because of the different frame types.

Data Link Layer Devices

The Data Link Layer deals with addressing on the local physical network. Bridges operate at the Data Link Layer. They use the Data Link Layer and its physical addressing to join several networks into a single network efficiently.

Functions

Bridges join two or more network segments together, forming a larger individual network. They function similarly to a repeater, except a bridge looks to see whether data it receives is destined for the same segment or another connected segment. If the data is destined for a computer on the same segment, the bridge does not pass it along. If that data is going to a computer on another segment, the bridge sends it along.

Bridges use a routing table to determine whether data is destined for the local network or not. On a bridge, the routing table contains MAC addresses. Each time the bridge receives data; it looks in its routing table to see whether or not the data is destined for a node on the local network. If it belongs to the local network, it does not forward the data. If it is not destined for the local network, it looks in the routing table to determine which physical network the destination address resides on, and sends the data out onto that network.

Bridges can not join dissimilar networks. If you have an Ethernet network and a Token Ring network, you cannot use a bridge; you must

40

40

use a router. However, a bridge can join networks that use the same frame type but different media, just like a repeater.

Relation to OSI Layer Functions

Bridges work with the MAC sublayer of the Data Link Layer. Remember that the Data Link Layer is concerned with communicating on the local network only. Bridges use information from the MAC sublayer to make decisions on whether a packet is destined for the same network or another network. The MAC address is used by bridges to determine first if the destination is local or not, then to choose which connected network it must go to.

Examples

Bridges are usually used to minimize network traffic. As a company’s network grows, it becomes busier and slower. An inexpensive way to minimize these growing pains is to segment the LAN using bridges.

Suppose the Marketing and Accounting departments described previously, linked via a repeater, continue to grow. Suddenly you, the network administrator, are receiving complaints from both departments that the network is very slow. A quick and simple way to fix this problem is to replace the repeater with a bridge. Now, anytime someone in the Marketing department accesses the Marketing server, their network traffic won’t cross over the Accounting department’s network. But if that same person wants to send email to accounting, that message will go right through.

Bridges are especially useful if many departments are connected together. If the manufacturing and shipping departments also want to be connected to Marketing and Accounting, repeaters are an inefficient way to add them in. Bridges can minimize traffic and allow the network to continue to grow, without “growing pains.”

Network Layer Devices

The Network Layer is concerned with network addressing for larger networks that consist of many physical networks, often with multiple paths between them. Routers operate at the Network Layer. They use the addressing information provided at the network level to join the many networks together to form an internetwork.

Functions

Routers divide larger networks into logically designed networks. Routers may seem a lot like bridges, but they are much smarter. Bridges cannot evaluate possible paths to the destination to determine the best route. This can result in inefficient use of network resources. Bridges also cannot use redundant paths. While two bridges can connect two networks, they risk sending packets in an endless loop

41

41

between the two networks. This behavior eventually saturates the network, rendering it unusable.

The drawback to a router’s inherent intelligence is their speed. Because they process so much information, routers tend to be slower than bridges.

Relation to OSI Layer Functions

Routers operate at the Network Layer of the OSI model. The Network Layer provides addressing for internetworks, and routers use this addressing information to determine how to pass along packets of data. Because routers operate at the Network Layer, they can link different physical network topologies.

Examples

Routers minimize traffic on internetworks, much like a bridge, but they are used to make large internetworks much more efficient. As data travels through an internetwork, it only knows its destination. It has no idea how to get to that destination. Routers look at the data’s destination and determine the next step. Routers can evaluate the best route to send the data along based on traffic and the speeds of various links.

Figure 4: Routers in a network

Figure 2-4 shows how routers send data between a workstation and a server across a relatively simply internetwork. When the workstation sends its data to the server, it first goes to Router A, then on to Router B. At this point, Router B must make a choice. There are two links to Router C and the network that the server resides on. One link is a fast T1, the other a comparatively slow modem link. Router B must decide which link to use given the current conditions, and send the data down that link. Router C receives the data and drops it onto the server’s local network.

42

42

Table 2-2 shows all the OSI layers and the various functions they perform.

Application Provides services on network, such as file/print, email, databases, etc.

Not an actual end-user application on a computer but provides services to applications.

Advertises available services to the network.

Presentation

Deals with syntax for communication between two computers.

Converts system data from the Application Layer to machine-independent format for the lower layers.

Receiver must convert machine-independent data to the local system’s format. This can include:

Bit-order translation – Either most significant digit (MSD) or least significant digit (LSD)

Byte-order translation – Determines which end of multi-byte values arrives first. Little endian takes least significant byte first, big endian takes most significant byte first.

Character code translation – Different binary schemes for characters (ASCII vs. EBCDIC)

File syntax translation – File formats that differ (Mac file forks vs. PC flat files)

Session

Manages dialogs between two computers by establishing, managing, and terminating communication.

There are three forms of dialogs:

· Simplex – One-way data transfers

Half-duplex –Two-way data transfers, but data can only flow in one direction at a time

Full-duplex – Two-way simultaneous data transfers

A session is a formal dialog between a requestor and provider and must have three phases:

Connection establishment – Requestor initiates service and communication rules are

43

43

agreed to.

Data transfer – Due to rules, each side knows what to do, operates efficiently, and detects errors.

Connection release – When session is done, dialog is terminated in an orderly fashion.

Transport

Can implement procedures to ensure reliable delivery of messages to destinations.

Allows upper layers to communicate with network while hiding network complexities.

Takes on some responsibility for connection services and interacts with Network Layer’s connection-oriented and connectionless services.

Segment sequencing – Connection-oriented service that resequences segments for reassembly.

Error control – Detects and takes care of corrupted, duplicated, or lost segments.

End-to-end flow control – Uses acknowledgements to manage flow between two devices. Some Transport Layer protocols can request retransmission of recent segments.

Network

Handles addressing and delivering packets on complex networks.

Uses logical network addresses to route packets.

Supports service addresses to specify a channel to a specific process on a destination.

Universally define service addresses are called well-known addresses.

Packet switching – Messages divided into smaller packets that contain addressing information, and can be sent through switches rapidly without being stored. Two types of packet switching:

Datagram – Each packet treated independently, and they can travel different paths getting to destination.

Virtual Circuit – A well-defined path through the network is negotiated, and remains in effect until communication is done. Network looks like a physical circuit, even though none exists.

There are two different modes of communicating across the network:

Connection-oriented – Error correction and flow control provided at internal nodes along path.

Connectionless – Internal nodes along path do not deal with error correction and flow

44

44

control.

When two networks are too different; a gateway is used to connect the networks.

Router and brouters operate at Network Layer.

Data Link

Assembles and disassembles frames for transmission and reception.

IEEE 802 breaks it down to two sublayers:

Media access control (MAC) – Allows multiple devices to share media.

Logical link control (LLC) – Starts and maintains links between devices.

Controls transmission method (synchronous/asynchronous) used to access media.

Maintains physical device addresses (MAC addresses).

Provides flow control and error control for single links between devices.

Flow control – Determines how much data to send to avoid overwhelming receiver.

Error control – Detects errors in received frames and requests retransmission.

Bridges operate at Data Link Layer.

Physical

Concerned with transmitting and receiving bits. Doesn’t define media, but defines:

Physical structure of network (physical topology)

Mechanical and electrical specifications for using medium

Bit encoding and timing

Repeaters operate at Physical Layer

Network Interface CardsNetwork interface cards provide a hardware interface between physical transmission media and the networked computer. They also provide a software interface that uses network protocols to format, transmit, and

45

45

receive information across the network. This chapter will focus on the hardware considerations for choosing a network interface card, as well as the software issues surrounding network drivers. It will also include practical information on installing and troubleshooting network interface cards.

Role of NICs

Network interface cards (NICs) are known by a variety of names including network adapters or cards, network adapter boards, and media access cards. Regardless of their name, they share a common set of functions in enabling computers to communicate across a network. Network interface cards are often defined by:

The type of Data Link protocol they support, such as an Ethernet adapter or a Token Ring adapter The type of media they connect to The data bus for which they were designed

In order to interact with the computer where it is installed, the network adapter, like any other peripheral device, must have a software driver installed. This driver allows the operating system and higher level protocols to control the functions of the adapter.

The NIC performs the following functions:

Translates data from the parallel data bus to a serial bit stream for transmission across the network. Formats packets of data in accordance with protocol. Transmits and receives data based on the hardware address of the card.

Translating from Parallel to Serial

The network adapter, like any device installed in a system, is attached to the CPU via a parallel data bus, such as an ISA or PCI bus. These parallel buses move data quickly between devices internal to the system. Network transmission, on the other hand, relies on a serial data stream in order to transmit messages. Therefore, all data coming into the network card from the bus must be transformed into a serial data stream before transmission.

Formatting Packets

Data transmission within a network interface card has two characteristics. The first of these is the transmission of binary bits across the network medium. A Physical Layer protocol defines the voltages and signal standards used to represent binary data on the network. This is built into the hardware on the card. In addition, the Data Link protocol supported by the card defines a packet structure for sending and receiving data. This structure includes fields for addressing and control information as well as the data. The card and its software driver implement this function.

46

46

Transmitting and Receiving Data

The primary function of the network adapter is to transmit and receive data between other network interface cards across the network media. It does this based on the fact that each card has an address, known as the MAC (media access control) address or hardware address. This address is generally built into the card by the manufacturer, and includes portions that identify the brand, as well as uniquely identifying the card itself. Each card is designed to pick out messages on the network destined for their hardware address as well as broadcast messages.

Network Connectivity

Expansion within a single network is called network connectivity. To expand a single network without breaking it into parts, one of the following devices can be used:

Repeaters

A repeater is probably one of the conceptually simplest devices on a network. Repeaters, although not necessary for functionality, enable your network to span a greater distance.

Function

All transmission media attenuate (weaken) the electromagnetic waves that travel through them. Attenuation, therefore limits the distance any medium can cover. Repeaters allow you to extend your network beyond the physical limits of the cabling. A repeater accomplishes this simply by repeating packets from one side of the wire to another while increasing or boosting the signal.

Repeaters fall in two categories: amplifiers and signal-regenerating repeaters. Amplifiers simply amplify the entire incoming signal. Unfortunately, they amplify both signal and noise. Signal-regenerating repeaters create an exact duplicate of incoming data by identifying it amidst the noise, reconstructing it, and retransmitting only the desired information. This reduces noise.

OSI Physical Layer

A repeater resides within the Physical Layer of the OSI Model, meaning that it does not look within the packets; the repeater simply receives packets on one side, and sends them on another side. A repeater never filters packets; all packets are always repeated.

Hubs

Hubs are one of the most important components of a network. They are the central location that all cabling must connect to in most topologies. (See Figure 10-1.)

47

47

Figure 1: A hub is connected to computers in a local network using cables

Role of Hubs in Topologies

Most network topologies can use for a hub in one way or another. The most prominent user of hubs is the 10BaseT topology. 10BaseT is entirely dependent on hubs for the infrastructure of the topology.

Passive

The function of a passive hub is simply to receive data from one port of the hub and send it out to the other ports. For example, an 8-port hub receives data from port 3 and then resends that data to ports 1, 2, 4, 5, 6, 7, and 8. It is as simple as that.

Active (Multi-Port Repeaters)

An active hub provides the same functionality of a passive hub with an additional feature. Active hubs repeat the data while resending it to all of the ports. By using active hubs you can increase the length of your network. It is important to remember that UTP (unshielded twisted pair) Category 5 cabling can be run a maximum of 100 meters. With an active hub, you can run Category 5 UTP 100 meters on each side of the hub.

3) Hybrid

A hybrid hub is a hub that can use many different types of cables in addition to UTP cabling. A hybrid hub is usually cabled using thinwire or thickwire Ethernet, which is discussed later in this chapter. Hybrid hubs are the most common type of hub. Hybrid hubs are used to interconnect hubs that are further than the 100-meter limitation of 10BaseT.

Bridges

A bridge is a network connectivity device that connects two different networks and makes them appear to be one network. The bridge filters

48

48

local traffic between the two networks and copies all other traffic to the other side of the bridge.

Network Segmentation

A bridge is a simple way to accomplish network segmentation. Placing a bridge between two different segments of the network decreases the amount of traffic on each of the local networks. Although this does accomplish network segmentation, most network administrators opt to use routers or switches, which are discussed later in the chapter.

Bridges segment the network by MAC addresses, as illustrated in Figure 10-2. When one of the workstations connected to Network 1 transmits a packet, the packet is copied across the bridge as long as the packet’s destination is not on Network 1. The bridges uses the bridge routing table to calculate which MAC addresses are on which network.

Figure 2: Bridges segment the network by MAC Address

Source Routing and Spanning Tree Bridges

Of the two primary types of bridges, source routing and spanning tree bridges, there is no major difference except for their applications. The end result of both types of bridges is the same, even though the mediums that they are used on are different. Spanning tree bridges are used to connect rings in a Token Ring network and source routing bridges are used for Ethernet networks. (See Figure 10-3.)

Building a Bridge Routing Table

Most bridges maintain their own bridge routing table dynamically and do not require that the administrator manage it unless he/she wishes to make a manual change to the table. To make a manual change to your bridge routing table, refer to the instructions that accompanied your bridge.

49

49

Figure 3: A bridge connects two local networks.

OSI Data Link Layer

A bridge is constantly tracking the destination MAC addresses of all packets that it receives. If the packets are determined to be ones that should cross the bridge, it passes them across the bridge. Since the only information the bridge knows about the packet is the MAC address of the destination, the bridge is said to reside in the Data Link Layer of the OSI Model.

Switches

Switches have become an increasingly important part of our networks today. As network usage increases, so do traffic problems. As a systems engineer, you will be faced with this problem on an almost continuous basis. A common solution to traffic problems is to implement switches.

Multi-Port Bridging

Switches also referred to as multi-port bridges, automatically determine the MAC addresses of the devices connected to each port of the switch. The switch then examines each packet it receives to find its destination MAC address. The switch then determines which port the packet is destined for and sends it out to that port only.

Network Performance Improvement with Switching

The primary benefit of implementing switching technology is that network performance will be improved a great deal. It is important to note that if you are not having traffic problems on your network, adding a switch will probably not change your network’s performance. If your network is having traffic problems, switching, when implemented properly, can greatly increase your performance.

Switching is a fairly involved process, as illustrated in Figure 10-4. Computer A transmits a packet to Computer C. The packet enters the

50

50

switch from Port 1 and then travels a direct route to Port 3. From Port 3 the packet is transmitted to Computer C. During this process, Computer B is unaware of the traffic between Computer A and C because there was a direct path within the switch and no shared bandwidth.

Figure 4: In a switch, data enters at Port 1 and is sent to Port 3 without sharing bandwidth with Computer B.

OSI Data Link Layer

Switches operate by knowing the destination MAC address and allowing the packet to use the direct route within the switch from the source port to the port that the device with the destination MAC address is connected to. The only information that a switch needs to operate is the MAC address, so the switch is said to reside in the Data Link Layer of the OSI Model.

Routers

Routers operate similarly to switches. The major difference is that a router is not quite as intelligent as a switch. Where a switch calculates which devices are connected to each port, a router receives packets from one side and determines if the destination is on the other side of the router. If the destination of the packet is on the other side of the router, the packet is forwarded. If the destination is not on the other side of the router, it is then forwarded to the next router.

Internetworks and Subnets

Rather than dealing with the MAC address of the destination of each packet, a router is concerned with the IP address of the packet (see Figure 10-5). The router determines if the packet’s destination is on one side of the router by using a subnet mask and the network address. The network address is the first IP address available on the subnet. The router then uses the logical AND operation to calculate which addresses it will allow to pass through. For example, a router

51

51

with a network address of 10.10.10.1 and a subnet mask of 255.255.254.0 allows all IP addresses between 10.10.10.1 and 10.10.11.255 to be routed. All other addresses are forwarded to the IP address defined as the default route. IPX is also capable of being routed in a similar fashion.

Figure 5: When routers are used to connect multiple networks, they only allow packets destined for the local network to enter the local network

Routing Protocols and Routing Tables

Routers keep a table of the routes to each possible destination. Most routers allow you to manually modify this routing table. Many older routers require you to follow their procedures and manually enter the routes to possible destinations. This usually consists of entering a network ID, subnet mask, and gateway address for each router port to be used.

Some of the more advanced routers incorporate Routing Information Protocol (RIP). RIP automatically calculates the quickest route to a destination and enters it into the routing table. RIP-equipped routers can automatically determine a new route to a destination if a router that it had previously been using is no longer available.

Brouters

A brouter is a hybrid of both a bridge and a router and has a connection to more than two networks. When the brouter receives a packet from one segment of the network, if it determines that the packet is not destined for an IP address located on the other side of the router, it is sent to the gateway address. If the destination IP address is connected to one of the other ports of the brouter, the packet is bridged to the other port instead of being routed. If a brouter determines that a packet received from one segment is not destined for a port of the brouter it will be routed. If the packet is destined for a port of the brouter, it will be bridged to that port.

Bridging vs. Routing Protocols

Routing protocols are quite different from bridging protocols. Bridging protocols are designed to combine packets from multiple physical networks and consolidate them into one virtual network. Routing protocols are designed to separate a physical network into multiple virtual networks.

The manner in which bridging and routing protocols operate also varies. A bridging protocol allows all traffic to cross that is not destined for the local network. A routing protocol allows traffic to cross that is destined for networks on the other side of the router.

52

52

OSI Network Layer

Routers and brouters operate by using the IP address to calculate if the packet should be routed. Since the IP address is needed, routers and brouters are said to reside in the Network Layer of the OSI Model.

Gateways

A gateway is a device that enables two dissimilar systems that have similar functions to communicate with each other.

Connecting Dissimilar Systems

Dissimilar systems are defined as two systems that have similar functions, but are unable to directly communicate with each other. For example, two Token Ring networks and Ethernet qualify under this definition.

In Figure 10-6, the PC on the left is able to use data and applications that are on the Mainframe on the right. As packets cross the gateway from the Mainframe to the PC, the Gateway converts them to a format that is understandable by the PC. The opposite operation occurs when the PC sends packets to the Mainframe. In this figure, PC and Mainframe are dissimilar systems.

Figure 6: Gateways connect two dissimilar systems and allow them to share data.

Wide Area Networks (WANs)One of the best known wide area networks (WANs) is the Internet. It grew out of ARPANET, a government defense research project for connecting sites in order to share radar data. The WAN technology that grew from there is different than traditional local area network (LAN) technology because it must allow for data transfer over a distance that is greater than the limitations of LAN physical media. For instance, the limit of 100BaseT over Category 5 cabling is 100 meters. This removes the possibility of connecting a New York office to a Los Angeles site via 100BaseT over Category 5 cabling.

A good way to think about WANs is to consider them to be the connections between two or more LANs. WANs are prevalent in business data communications if that business encompasses more than one office. Even if a business does not have more than one office, it may need to communicate with other businesses in order to conduct daily operations. This may range from a business connecting to a bank

53

53

in order to have online accounting, to the bank providing data and applications to customers via kiosk ATMs.

Many businesses are using the Internet to conduct part or all of their daily business. The Internet is the world’s largest WAN. With the Internet’s popularity increasing by leaps and bounds, being able to connect to, use, and troubleshoot an Internet connection may soon be a requirement for all network systems engineers.

Carrier Services and Modems

Imagine dialing into the Internet. What has to take place in order to get from a PC to the Internet?

First, there has to be an Internet access point. For people who dial in, this is the computer that is connected to the modem they are connecting with. It is also known as a network node or a host or a remote access server, because of its place on the Internet and its function. The generic term for both the PC and for the remote access server is data terminal equipment (DTE).

Second, there has to be communications equipment at both the Internet access point and at the PC. This can be an ISDN codec, an analog modem, or WAN equipment such as a CSU/DSU. The communications hardware is generically called data communications equipment or data circuit-terminating equipment (DCE). Figure 11-1 illustrates the use of DTEs and DCEs to connect to the Internet.

Figure 1: Connecting to the Internet using DTEs and DCEs

Third, the DTE communicates via the DCE to the remote DTE via its DCE. In order to communicate, there must be a common language. This language is considered a carrier signal when using an analog line. A carrier signal is an analog signal whose characteristics—the frequency, the amplitude, and/or the phase—have been modulated to represent data. Analog signals continuously vary in a wavelike course. Digital signals, on the other hand, are either positive, zero, or negative at a certain voltage; there is no smooth transition between a positive and negative electromagnetic signal. Figure 11-2 illustrates these differences.

54

54

Figure 2: Analog and digital signals

Fourth, the carrier signal must have some path to travel between the DCEs. Transmission media, or the carrier, provides the path. A Carrier Service provides the transmission media. Carrier services are traditionally telephone companies. They offer their existing telephone wires as the path between the DCEs. Keep in mind, though, that a carrier service can also be a satellite service using wireless transmission, a cable company using their existing cable network, or even a network using another media perhaps installed just for data transmission.

When using an analog line, the DCE must translate the digital data to an analog signal. The digital signal must be modulated before it is sent over the wire. When it gets to the remote DCE, the analog signal must be demodulated into a digital signal that the remote DTE will understand. The word modem, therefore, is shorthand for Modulator Demodulator.

Analog Carriers

Analog carriers are traditionally telephone company lines, or the Public Switched Telephone Network (PSTN). Carrier services, other than the telephone company, that offer data networks usually create digital networks also called public data networks (PDNs). The types of WAN connections that may use the PSTN include:

Dial-up lines Leased lines Switched 56 T-1 to T-3 lines (known as T-carrier lines) ISDN

Of these WAN connection types, the dial-up lines are usually analog. Leased lines, switched 56, T-carrier, and ISDN connections are usually digital.

55

55

Dial-up or Switched lines

Why are dial-up lines also switched lines? Because the public telephone network is a switched service, and dial-up lines use those wires and circuits to create a connection. The telephone network is set up so that a signal travels to the local Central Office, one of which is usually located somewhere within a few miles of any telephone. A pair of wires that create the electrical current needed for stable communication wires each phone to the central office. The area code and the exchange of the phone number identify which central office is used. The exchange is the first three numbers of the seven-digit phone number.

A dial-up line is a communications circuit that is established by a circuit-switched connection using PSTN.

Leased Lines

A leased line is similar to a dial-up line. However, it is reserved by the carrier service for private use by the person or business leasing the line. Leased lines are dedicated, as opposed to switched, and always follow the same path. Leased lines use a digital signaling scheme, rather than analog. This makes them capable of greater speeds than analog systems. Leased lines generally incorporate high-grade copper or fiber-optic media. This makes them capable of higher speeds than ISDN, which typically runs over standard-grade telephone wire.

Although there can be more complicated configurations, the simplest leased line is a permanent line leased from the carrier service (usually the telephone company) between two sites that need to share data. A customer service unit/data service unit (CSU/DSU), which is similar to a modem, is placed at each end of the leased line. The CSU/DSU, which is the DCE, is fed data from the DTE, which then transmits it across the leased line to its CSU/DSU partner, which then decodes the data and forwards it to its attached DTE.

Leased lines can use different framing protocols. Framing protocols occur at the Data Link Layer of the OSI protocol stack. Point-to-Point Protocol (PPP), discussed later in this chapter, is a typical leased line protocol, as is High-level Data Link Control (HDLC), which is not discussed.

The first disadvantage of leased lines is cost. The line is being paid for all the time, whether or not data may be flowing between the two sites, and whether or not full bandwidth is being utilized. There is a certain point of bandwidth utilization called the break-even point. It is the point midway between where a dial-up line is cheaper than a leased line and where a leased line is cheaper than a dial-up line. Figure 11-3 illustrates this cost comparison.

56

56

 

Figure 3: Comparing the costs of a leased line vs. a dial-up line

The second disadvantage of leased lines is scalability. Each link is direct between two sites. In order to increase connected sites, there have to be multiple CSU/DSUs, multiple router ports, and multiple leased lines. This lack of scalability also affects costs.

Digital Carriers

Leased lines are an example of digital carriers. There are several categories related to bandwidth. The smallest size is 64 Kbps, and is called a DS-0 line. DS-0 stands for Digital Signaling 0 and is the base unit for all greater sized digital lines. Think of a DS-0 as a framing specification used in transmitting digital signals over a single channel at 64 Kbps on a T1 facility, where the T1 facility would have a total of 24 channels.

Table 11-1 lists the T-carrier rates and their characteristics.

Name

Digital Signal frame

# Of channels

Bandwidth Size

Media

64 DS-0 1 64 Kbps CopperT-1 DS-1 24 1.544 Mbps CopperT-2 DS-2 96 6.312 Mbps CopperT-3 DS-3 672 44.736

MbpsFiber or other high-speed media

T-4 DS-4 4032 274.76 Mbps

Fiber or other high-speed media

57

57

Table 1: T-Carrier Rates and Media

T-1

AT&T developed T-1 for digital transmission. It uses time-division multiplexing to allow 24 separate channels to combine for a total bandwidth of 1.544 Mbps. Each channel is 8 bits wide. T-1 adds a synchronization bit every 193 bits. This is the clocking mechanism for the data flow. T-1 is not dependent on the physical media used.

Since it uses time-division multiplexing (TDM), the T-1 carrier signal is sampled and interleaved, and then converted into a digital data stream. TDM is a synchronous system that interleaves fragments of slower channels into a single faster channel. TDM is the only system that can be used on a baseband line. Figure 11-4 illustrates a four-channel TDM system.

Figure 4: Four-channel time-division multiplex system

T-1 is similar to the European standard E-1, which is same in structure but maxes out at 2.108 Mbps.

A common configuration for a leased line is called fractional T-1. This is a combination of multiple 64-Kbps channels within the 24 available channels of a T-1 line. Most common are 128-Kbps or two channels, and 256-Kbps or four channels.

58

58

T-3

T-3 is larger than T-1. As a result, it is more expensive. T-3 transmits DS-3-formatted data at 44.736 Mbps over optical fiber or microwave lines. The DS-3 is the framing specification. T-3 is the equivalent of 672 DS-0 channels.

Switched 56

Switched 56 is the low end of WAN point-to-point services. Switched 56 is a something of a misnomer. Although it is commonly sold as 56 Kbps, some areas offer 64 Kbps, although that is more common in Europe. The reason that US services offer 56 Kbps is due to the management overhead of the line. It is also a digital service.

Switched 56 is called that, since it is an on-demand dial-up line. It is still based on a T-1 channelized system. When the customer dials up the line, one of the channels is switched into the customer’s link. This means that the costs are based on the actual usage. Most carrier services offer a dedicated 56-Kbps service, as well. Switched 56 is an excellent backup link, providing fault tolerance for other higher speed lines.

ISDN

ISDN means Integrated Services Digital Network. Originally created by the International Telecommunications Union Telecommunication Standardization Sector (ITU-T), ISDN is a project to upgrade the existing PSTN to be able to run digital services and provide digital connectivity between video, terminals, telephones, computers, voice mail, etc. Because it uses the existing telephone network, ISDN uses copper wire.

ISDN components are illustrated in Figure 11-5 and include:

TE1 (terminal equipment type 1) ISDN terminals TE2 (terminal equipment type 2) Terminals that predate ISDN. NT1 (network termination type 1) Equipment that connects the subscription four wires to the two-wire local loop, is provided by the customer in the US, but provided by the carrier in other countries. NT2 (network termination type 2) Performs protocol functions of the OSI protocol stack Data Link and Network Layers. TA (terminal adapter) Used with a TE2 in order to adapt it to ISDN, and can be either internal or external.

59

59

Figure 5: Integrated Services Digital Network

The ISDN structure consists of the following digital channels, also called bit pipes:

A analog telephone, 4 kHz B digital data, 64 Kbps C digital out-of-band, 8 or 16 Kbps D digital out of band, 16 or 64 Kbps with 3 sub channels: s for signaling, t for telemetry, and p for packet data E digital channel for internal ISDN signaling, 64 Kbps H digital channel at 384, 1536, or 1920 Kbps

The ITU-T ISDN project defined three standard channel combination services:

Basic Rate (BR) 2 64-Kbps B channels and 1 16-Kbps D channel Primary Rate (PR) 1 64-Kbps D channel, with 23 B channels or 30 B channels in Europe Hybrid 1 A channel and 1 C channel

The Basic Rate is the most common configuration for end users connecting to a Remote Access Service or Internet connection. The Primary Rate is more common for business use in site-to-site communications.

Table 11-2 lists the carrier types and their access method and bandwidth.

Carrier Type Switched or Dedicated

Bandwidth

Dial-up lines Switched Up to 56 Kbps

60

60

Leased lines Dedicated 56 Kbps to 1.544 Mbps

Switched 56 Switched 56 KbpsT-1 Dedicated 1.544 MbpsT-3 Dedicated 44.736 MbpsISDN – Basic Rate Switched 2 64-Kbps B channels

(with 1 16-Kbps D channel) reaching a general rate of 128 Kbps

ISDN – Primary Rate Switched 23 64-Kbps B channels and 1 64-Kbps D channel = 1.544 Mbps.

Table 2: Carriers, Access, and Bandwidth

Modems

Modems are the communications equipment used to modulate digital data into an analog signal to be sent across a standard telephone wire. In a modulation/demodulation system, the CPU feeds digital data, in the form of 1s and 0s, into the modem. The modem analyzes the data and transforms the digital signals to analog signals, which can be sent over a phone line. Another modem then receives these signals, converts them back into digital data, and forwards that data to the receiving CPU.

Functions

Modems do have different sub functions, beyond the modulation/demodulation function. They can:

Compress data Handle error correction Control the flow of data Buffer the data to prevent scrambling

Data compression is used within a modem to send the same amount of data using fewer bits. Although data compression algorithms are somewhat complicated, they are akin to shorthand, where fewer pen strokes equal the same amount of words.

Error detection is the method by which modems verify that the information sent to them has been undamaged during the transfer. In order to correct errors, they must be detected first.

Error-correcting modems break incoming data into frames. Then a checksum is attached to each frame and the formatted data is

61

61

forwarded to the receiving modem. The receiving modem verifies the checksum to ensure it matches the information sent. If not, the receiving modem notifies the sending modem and then the entire frame is resent. Though error correction slow down data transfer on noisy lines, it provides greater reliability. As with data compression, for error correction to be used, both modems must support the same error correction standard.

Flow control was a necessity once modems with differing capabilities needed to communicate with each other. Since one modem in a connection was capable of sending data much faster than the other could receive, flow control was created to be able to pause the sending modem while the receiving modem caught up. The varieties of flow control are:

Software XON/XOFF flow control Hardware request to send/clear to send (RTS/CTS) – flow control

With software flow control, when the receiving modem indicates that a pause is needed, it sends a certain character, usually Ctrl-S. Since the Ctrl character is typically sent, these commands are sometimes called control indicators. When the receiving modem is ready for more data, it sends another control indicator, such as Ctrl-q. The advantage of software flow control is that it can use a serial cable with only three wires. One disadvantage is that line noise can sometimes inadvertently create control indicators, pausing or restarting transmissions at inappropriate times. Another disadvantage lies in the fact that since binary files contains control characters, they should never be sent using software flow control.

Hardware, or RTS/CTS, flow control bases its pause and resume features on whether the RTS or CTS wires in the modem cable send a bit of data. In the case of an internal modem, this feature is within the modem hardware itself. Hardware flow control tends to be faster and more reliable than software flow control.

Universal Asynchronous Receiver/Transmitters (UARTs) are the hardware pieces designed for the computer to send information to a serial device. A modem is a serial device, as well as a mouse, and other input/output devices. The UART is an integrated circuit (chip), which adapts parallel input into serial output. Parallel data transmission occurs across several pins at one time—the data travels “in parallel.” Serial data transmission occurs in a series across one pin (or two for duplex). The CPU feeds the data into the UART’s buffers. The buffers use first in, first out (FIFO) scheduling so that the first data to enter the buffer is the first to leave. Without FIFO, information would be scrambled when forwarded through the modem.

62

62

National Semiconductor created the first UART called INS-8250. This was upgraded to a faster 16450. Both of these UARTs had a one-byte buffer, which meant that if information was traveling too fast for the CPU to handle, it was overwritten. Because of today’s faster modem speeds, the UART was upgraded to 16550A. The 16550 have a 16-byte buffer, so that a busy CPU can catch up after dealing with other tasks.

Internal vs. External

An internal modem and an external modem perform the same function. An internal modem is either built into the motherboard of a PC or laptop, or the internal modem is an additional adapter card within the computer. If the computer being used is older, say a 386, the serial port UART is most likely the 16450, which means that the PC can only handle up to a 2400-bps external modem connected to that serial port. A UART is built into an internal modem, so using a faster modem requires an internal modem, or upgraded serial port card. Another advantage to internal modems is that they are cheaper, since there are no case or indicator lights that have to be included in the product.

The advantages of using an external modem lie in the ability to use the external serial port, and not have to use up an additional interrupt within the PC. An external modem is easily installed, and can be transferred between PCs with little or no trouble. If an external modem gets “stuck” in communications, perhaps due to a software flow control error; it is easily reset since it has its own on/off button. Rebooting the PC is the only way to reset an internal modem. An external modem usually has indicator lights that can help troubleshoot communication problems.

Connectors: RJ-11 and Serial

A connector is the part of a cable that plugs into a port or interface in order to connect one hardware device to another. Connectors are either males where they consist of one or more exposed pins, or female where they contain openings into which the male connector can be inserted.

A standard PC serial port is called an RS-232. The Electronics Industry Association (EIA) developed that standard for serial communication. In an RS-232 serial port, one pin is used for transmitting data and another for receiving data. There are other pins used to establish and maintain communications between the two serial devices. Standard serial connectors come in two sizes: 25 pins or 9 pins. Each pin represents a type of data signal to be sent. Because the type of data is restricted to certain pins, a cable must be wired so that it transmits or receives the same data across that pin as is expected.

63

63

Table 11-3 lists the RS-232 pin assignments.

9-Pin Connector

25-Pin Connector

Symbol

Signal sent on that pin

Input/Output data

1 8 DCD Data carrier detect Input2 3 RX Receive data Input3 2 TX Transmit data Output4 20 DTR Data terminal ready Output5 7   Signal ground  6 6 DSR Data set ready Input7 4 RTS Request to send Output8 5 CTS Clear to send Input9 22 RI Ring indicator Input- 1   Protective ground  - 9   Transmit current loop

+Output

- 11   Transmit current loop - Output- 18   Receive current loop + Input- 23   Data signal rate

indicatorInput/Output

- 25   Receive current loop - Input

Table 3: RS-232 Pin Assignments

As long as the basic 9-pin signals are maintained, serial cables can be wired that have a 25-pin connector at one end and a 9-pin connector at the other.

The RS-232 cable connectors are meant for communication between two serial port devices: a modem and a serial port. The connector between the modem and the telephone line is different. It is usually an RJ-11 connector, although more often upgraded telephone wiring is now using an eight-wire RJ-45 female connector that is backwards compatible for a male RJ-11 connector. This is being done in anticipation for further services to become available via the telephone network.

The RJ-11 connector is shorthand for Registered Jack-11. It is either a four-wire or six-wire connector used primarily to connect telephone equipment in the United States. Modems connect with an RJ-11 cable into the RJ-11 female port that leads to the PSTN. In the standard six-wire configuration, the wires are configured as

Pin Function

1 CTRL In

64

64

2 Ground

3 Data Out

4 Ground

5 Data In

6 CTRL Out

Modem Standards- Hayes and ITU V Standards

The ITU-T is responsible for the V series of modem standards for modulation, data transfer, and data compression protocols. In the early days of modems, Bell created modulation standards, such as Bell 103 and Bell 212A, that were prevalent in the United States. The ITU-T (which was the CCITT at the time) was responsible for creating international standards.

Some V series modem type descriptions are followed by a “bis” or “ter.” That suffix represents a secondary or tertiary V. series standard. For instance when comparing a V.32 modem with a V.32bis modem, the bits per second rate is significantly different; V.32 = 9600bps and V.32bis = 14,400bps, even though the base modem standard is the same.

The MicroCom Networking Protocol (MNP) is also prevalent in modems. Rather than a series, MNP offers various classes of standards offering error detection and error correction abilities. MNP class 5 has been popular in the past because it offers data compression, effectively doubling the modem bps rate.

Standards created by an organization are called de jure standards. Standards that are adopted due to their prevalence, convenience, or the lack of other standards are called de facto standards.

The Hayes modem standards are de facto standards. In the early days of modems, Hayes had the largest presence in the industry. Any standards that they developed, they applied to all their modems. As a result, many software programs were written that accommodated the Hayes modem standards, such as AT commands. AT commands are the control characters sent to a Hayes standards-based modem. In order to compete, other manufacturers had to make their modems compatible with the software programs by incorporating AT commands, as well as making sure their modems could communicate with the Hayes modems, by utilizing the Hayes software XON/XOFF flow control ability.

V Standard

Transmission Rate

V.22 1200 bpsV.22 bis 2400 bpsV.32 4800–9600 bpsV.32 bis 4800–14400

65

65

bpsV.34 2400–28800

bps*V.fc 2400–28800

bps*V.fast 2400–28800

bps*V.fc and V.fast are not ITU-T standards, they are proprietary.

 

Table 4: V Standards and Transmission Rates

Baud Rate vs. BPS and Compression

Baud stands for bits of actual usable data. Baud represents the number of carrier signal level changes per second within the modem. Each signal level contains one or more bits of information. BPS is the acronym for Bits Per Second, where bit is a unit of data representing a 0 or 1, and is the rate of data flow across the telephone wire. Since baud rate and BPS used to be the same number, they can be confused. But the baud rate can be a different number than BPS if the number of signal level changes within a modem is different than the number of bits transmitted.

In order to go beyond a 2400 bps rate—which is approximately the bit rate maximum of the telephone line—modems had to be able to encode more than one bit of data in every signal transition. The maximum bit rate of the telephone wire can be higher, but usually 2400 bps is about right. By encoding more bits of data per signal transition, the modem may work at 2400 baud, but be transmitting 28,800 bits per second. This is a form of modem data compression called DCE-to-DCE compression.

There are two types of compression: DCE-to-DCE compression and DTE-to-DTE compression. DCE-to-DCE compression takes place between two modems. Both modems must support modem compression in order for any compression to take place. In modem compression standard v.42 bis, a 4:1 ratio of compression is achievable. In modem compression standard MNP5, a 2:1 ratio of compression is available. Data travels from the DTE to the DCE where it is compressed and sent to the remote DCE where it is expanded and then forwarded to the receiving DTE.

66

66

DTE-to-DTE compression occurs between the remote access server and the remote node. Both the remote access server and the remote access client must support DTE-to-DTE compression. Here, the data is compressed at the DTE, and then sent to the DCE where it is forwarded to the remote DCE and then to the remote DTE, where it is expanded. Note that in Windows NT Remote Access Service, a higher data compression ratio of 8:1 is achievable if both RAS compression (DTE-to-DTE) and v.42 bis compression (DCE-to-DCE) are available and enabled.

Asynchronous vs. Synchronous

Asynchronous and synchronous are two different techniques for data transmission.

In asynchronous transmission, data is coded into a series of pulses, including a start bit and a stop bit. A start bit is sent by the sending modem to inform the receiving modem that a character is to be sent. The character is then sent, followed by a stop bit designating that the transfer of that bit is complete.

Asynchronous transmission manages the data flow by transmitting each character separately, with its own separate synchronization information. A start bit and a stop bit frames each character, which consists of a single byte of data, and is sent as a single transmission string. A byte of data equals eight bits. The start and stop bits synchronize the data transmission between the sending and receiving DCEs.

Asynchronous transmission is a mature, simple technology, using common and inexpensive hardware (modems). It has the disadvantage of having a high overhead, comprising between 1/5 and 1/3 of the bandwidth used. As a result, data transfers are slow. Also, asynchronous transmissions are subject to errors due to the nature of the legacy telephone wires that are typically used. Figure 11-6 illustrates how a single character is transmitted using asynchronous transmission.

Figure 6: Asynchronous transmission of a single character

In synchronous data transmission, data is sent via a bit stream, which means that a group of characters is sent at once without start and stop

67

67

bits interrupting them. For synchronous communication, the groups of characters are gathered into a buffer of the DCE, where the DCE formats the data to be sent as a stream. To prevent garbling, synchronous DCEs must be in precise synchronization. They accomplish this by sending control codes, called synchronization, or SYN characters. Once the DCEs are in synchronization, they transfer the bit stream of data.

Synchronous transmission sends both text and binary data the same as asynchronous, but it uses bandwidth much more efficiently since it can send large blocks of data instead of a single character at a time. In order to be able to transmit data in blocks, there must be some method of synchronizing the transmission between the sending and receiving DCEs without framing each character. Either the synchronizing signal can be contained in the frame around the data, or the synchronizing signal can be constantly transmitted separately. The SYN character is one method of accomplishing this.

Synchronous transmission is more efficient and much faster than asynchronous transmission. Due to the fact that it is newer and more complex, the hardware tends to be more expensive than asynchronous. Figure 11-7 illustrates two synchronous transmission modes: one with SYN characters interleave with the data and one with SYN characters transmitted separately.

Figure 7: Two synchronous transmission modes: interleaved and separate SYN clocking

68

68

Analog vs. Digital Modems

As discussed previously, an analog signal refers to information being presented continuously, while digital refers to data defined in individual steps. Analog information is usually in the form of a sound wave. As a result of the continuity, any interference or noise on the line can interrupt, change, or damage the signal being sent. Digital data is less affected by such interference.

As discussed previously, modem stands for modulator/demodulator. The function of the modem is to modulate digital signals into analog signals in order to send them out over an analog line, as well as to demodulate analog signals into digital signals in order to feed them to the CPU. It is always considered a DCE (Data Communications Equipment or Data Circuit-Terminating Equipment). Another term to be aware of is the codec, which stands for coder/decoder. A codec is somewhat the opposite of a modem. It changes data into a digital signal to go across a wire, and decodes it back to analog. A time-division multiplexor is an example of a codec. A codec is sometimes be referred to as a digital modem.

PPP and SLIP

Point-to-Point Protocol (PPP) and Serial Line Internet Protocol (SLIP) are most commonly used over analog lines, but also over ISDN and dedicated high-speed lines, to provide a remote node-to-network connection. Some routers use these protocols to connect two networks from router to router.

SLIP is a legacy UNIX communications standard. It provides a remote node connection between a workstation and a network or in router-to-router network connections. SLIP does not have the encryption capability of PPP. Neither does it automatically negotiate the connection when connecting to a network. Instead, user intervention is required. SLIP works at the Physical Layer of the OSI stack, so does not provide either error control or security.

Windows NT RAS (Remote Access Service) can be configured as a client for SLIP, but cannot be a server for SLIP.

PPP is the successor to SLIP. Again, it provides remote node-to-network or router-to-router connections over telephone lines and modems, ISDN, and high-speed links. PPP improved on SLIP by functioning at both the Physical and Data Link Layers, thereby being capable of providing error control, security, dynamic IP addressing, and support for multiple protocols. Within the Data Link Layer, the MAC (media access control) portion handles the physical addressing of the device, and the LLC (logical link control) portion handles error control for

69

69

connection services. Figure 11-8 illustrates where PPP and SLIP fit in the OSI protocol stack.

Figure 8: Where PPP and SLIP fit with the OSI protocol stack

The PPP frame consists of 6 fields: Flag, Address, Control, Protocol, Data, and Frame Check Sequence. The Flag field is one byte 01111110, indicating the beginning or ending of a frame. The Address field is a single byte 11111111, which is the standard broadcast address since PPP does not allocate specific station addresses. The Control field is a single byte 00000011 that sets up transmission of data in an unsequenced frame. The Protocol field is two bytes identifying the protocol. The Data field is a variable-length datagram for the appropriate protocol of the data being sent. Maximum size for the Data field is usually 1500 bytes, but the Data field can be a different size depending on the manufacturer. The Frame Check Sequence field, used for error handling, is either two bytes, or four bytes if set up for better error detection.

Figure11-9 shows the format of a PPP frame.

70

70

Asynchronous Protocols

Asynchronous protocols generally use digital signals transmitted without precise clocking. Hence, the term asynchronous. The signals may have differing frequencies or phase relationships, such as video or data. The data variation and the lack of clocking forces the need for another method of control in order to manage errors.

Traditional asynchronous transmissions, such as modem-to-modem dialogs over analog lines, encapsulate each character with start and stop bits in order to control the transmission. This is a tremendous overhead, making poor use of bandwidth.

Newer asynchronous protocols have different methods of data transmission in order to increase speed and reduce overhead. For instance, ATM utilizes a cell-switching technique in which a fixed 53-byte cell length and cell relay reduces transit delays. ATM was developed to transmit voice, video, or data easily over high speeds.

Tunneling and Virtual Private Networks

Tunneling a protocol refers to the ability to route a protocol, such as IPX or AppleTalk, over a network that uses a different protocol, such as TCP/IP. The result is a connection between the two networks that uses the network’s native protocol, be that IPX or AppleTalk, but which is encapsulated and transmitted in a TCP/IP packet format. Note that IP traffic can also be tunneled through an IPX or AppleTalk network, and AppleTalk can be tunneled through an IPX network and vice versa as long as the appropriate tunneling software is available and configured to format the data.

71

71

Figure 16: IPX protocol tunneled through an IP network

In Figure 11-16, the network workstations are able to communicate using IPX, even though the link connecting the two networks uses only TCP/IP. This is accomplished through the tunneling software, which in this case, is running on the routers since it encapsulates the IPX frames in IP packets. The data leaves the workstation with the IPX destination address of the workstation on the other network. The path in IPX is known to be through the router, so the IPX data arrives there. The router knows that it is forwarding this data over an IP network, and adds an IP encapsulation packet using the destination router as the final IP destination address. The IP network reads the IP header and knows to send the data to the destination router. Once at the destination router, the IP encapsulation is discarded and the IPX frame is then forwarded through the network using the IPX destination address to the receiving workstation.

In response to public demand, Microsoft created Point-to-Point Tunneling Protocol (PPTP). This protocol tunnels the Point-to-Point Protocol (PPP) over an IP network to create a network connection. The primary use for this capability is to use the Internet as a network connection.

Like a standard tunnel system, PPTP encapsulates packets of information (IP, IPX, or NetBEUI) within IP packets for

72

72

transmission through the Internet. At the destination, the IP packet encapsulation is discarded, and the original packets are forwarded to their appropriate destinations. Both encryption and authentication are used in PPTP. Encryption of the transmitted data protects the data. Authentication is used to verify the identity of the user in order to grant access to network resources.

Once connected via PPTP, a remote user has a virtual connection to the network. It is transparent in that the end user may use network resources just as if that user were connected directly to the network. The use of PPTP, in effect, creates a Virtual Private Network (VPN). This is sometimes referred to as an ExtraNet, and is a form of an Extranet, but is not limited to WWW or FTP applications. Unlike the Internet, a VPN is not wide open, even though it uses the Internet as a backbone network. It is, in fact, virtually private. VPN can be accomplished using tunneling protocols, since the data is encapsulated the way it is, it is secure.

Protocols Appropriate to Different Server Types

Depending on the server that is providing remote access services, the protocol will vary. Generally, the two types of servers traditionally used by an Internet Service Provider (ISP) are UNIX servers or Windows NT Servers. It is possible that other servers are used, as well.

UNIX SLIP Servers

Serial Line Internet Protocol (SLIP) is a legacy remote node protocol offered by UNIX servers. Because of its prevalence, SLIP is still used. SLIP works at the Physical Layer only of the TCP/IP protocol stack.

Using Dial-Up Networking in the current version, or the RAS client in older versions, Windows NT can connect to SLIP servers as a client only. Neither Windows NT Server, nor Windows NT Workstation can be configured as a SLIP server.

Windows NT PPP Servers

Windows NT Remote Access Service offers PPP as the default connection type for remote nodes. PPP is protocol independent and can be used with any of Windows NT’s native protocols: NetBEUI, NWLink, and TCP/IP. PPP is also the default client type for Dial-Up Networking and for the older RAS client in older versions of Windows NT.

73

73

Protocols Common to Internet Service Providers

Internet Service Providers (ISPs) commonly use PPP as the protocol for dial-up connections. They may also use SLIP for dial-up connections if the ISP’s servers are UNIX based.

In order to create a connection to an ISP using Windows 95 or Windows NT, TCP/IP and Dial-Up Networking/Remote Access Service client should be installed.

A local area network can connect to an ISP using a leased line or other high-speed access connection type. In that scenario, TCP/IP is routed directly over the line to the Internet.

Both Windows NT and Windows 95 support all the protocols needed to connect to an ISP, including their native 32-bit implementations of TCP/IP, and PPP or the SLIP client. In addition, both operating systems offer basic FTP and Telnet clients, which are used to download files and access Internet Telnet servers.

Table 11-5 lists the common server types for asynchronous protocol types.

Network Operating System

Function Protocol

UNIX Server SLIPUNIX Client SLIPWindows NT Server PPPWindows NT Client PPPWindows NT Client SLIP

Table 5: Matching Asynchronous Protocols and Common Server Types

74

74