Computer Networks: LAN, MAN and WAN | Networking | Computers

This article throws light upon the three main computer networks. The networks are: 1. LAN 2. MAN 3. WAN.

Computer Network # 1. Local Area Networks (LAN):

These networks are generally referred to as LANs. In fact, the acronyms LAN for Local Area Network and WAN for Wide Area Network have become so popular that many computer users would be quite nonplussed if one talked to them about “Local Area Networks” and “Wide Area Networks” instead of about LAN and WAN.

LANs are private networks, within a single building or a campus, where distances of not more than about 2 to 3 kilometers between different computers, are involved. Obviously, this distance limitation is placed so that certain characteristics can be highlighted and some specific technologies can be utilised.

The three main distinguishing characteristics of LANs usually are:

1. Geographical size of the area covered by the network,

2. Transmission technology,

3. Topology.

LANs are limited in size, so that the technology selected and the costs can be bound by the distances involved. This limitation in distances also has a bearing on the transmission technology and topology to be utilised in setting up the local area network. It also simplifies the process of the management of the network.

The transmission technology used usually consists of a single cable to which all the computers are connected. This technology usually makes the design and usage of the network very simple. LANs have low delays and usually operate at speeds of 10 Mbps to 100 Mbps (the measure Mbps refers to megabits per second as opposed to MBps or MB/sec, which refers to Megabytes per second).

They are also not error prone and in a LAN one may expect very few errors. The topologies used are also usually very simple. LANs invariably use either a bus topology or a ring topology. These are shown in Figs. 7.1 and 7.2.

While the use of bus or ring topologies is not essential in LANs, normal LANs generally use Ethernet cards. This invariably makes the use of bus topology mandatory, since Ethernet is designed for bus topology. Similarly, the token ring, a standard advocated originally by IBM is also frequently used in LANs.

This obviously implies that LANs, in general, are broadcast networks. A LAN can also be designed using point-to-point lines. But these are rare in LANs. In fact, if a LAN uses point-to-point connectivity, then it is actually a small replica of a WAN or Wide Area Network.

IEEE 802.3 refers to protocols and cabling for LANs including Ethernet. IEEE 802.4 provides the standards for the Token Bus, 802.5 for the Token Ring and 802.6 for DQDB for Metropolitan Area Networks.

Switched LAN (IEEE 802.3):

As clients are added to a LAN, its starts to get saturated, creating problems for the network. This problem could obviously be solved by increasing the bandwidth of the network from say, 10 Mbps to 100 Mbps, if that is possible. But this would imply changing the 10 Mbps adapter cards on all the computers in the network—a somewhat expensive proposition. However, this can be done by also using a switched LAN.

Usually called a switched Ethernet, it consists of a switch containing several plug-in line cards. Their number may vary, but 16 cards seem to be quite common. Each card can have up to 8 connectors. Each connector has a connection to a host computer. When a machine wants to send a frame, it sends the frame to the switch.

The card checks to see whether the message is intended for one of the machines on the card. If not, it sends it back via a high-speed “backplane“. In this way, the function of sending the message is distributed by the switch.

Token Bus (IEEE 802.4):

The Token Bus is another scheme for networking that was developed to take care of the basic weakness of the 802.3 Ethernet standard. This basic problem is the fact that there are no priorities between frames.

Thus, in the worst case scenario, an important message may get held up for a long time while some of the unimportant messages may get transmitted. In a token ring, the attached stations take turns to send messages.

Thus, if there are n stations attached to the ring and each takes time t seconds to send a message, no message can be delayed beyond nt seconds. There is, of course, the disadvantage that in case of the ring is broken, the entire network goes down. Furthermore, a ring is unsuitable as a topology in, say, an assembly line.

As a result, the token bus takes advantage of the concept of the linear broadcast, while still maintaining the advantage of the token ring. Logically, each station is attached in a ring, although not physically.

The stations are numbered, with each station being aware of the station before it and station after it. When a station has completed despatching its message, it passes a “token“, which is a special frame, to its neighbour (left or right, depending on the initial logic).

Only the holder of the token is permitted to send a message. Since the despatching of messages is done in this organised manner, there are no collisions. Since the arrangement of the token bus is a broadcast arrangement, the message sent by one station is received by all the stations attached to the ring, but it retains only those that are addressed to it.

Physically, the token bus is a linear cable onto which the stations are attached. Logically, the stations are organised into a ring. Each station is aware of the addresses of the station to its immediate right, and the station to its immediate left. When the ring is initialized the highest-numbered station may send the first frame. It then passes on a token, which is a special frame to its neighbour.

The token advances around this logical ring with only the token holder permitted to transmit frames. Since only one station at a time is permitted to transmit frames, there is no possibility of collision of frames. The physical order in which the stations are connected to the main cable is not important.

The cable being a broadcast medium, each station receives each frame, discarding the ones not addressed to it.

When a station passes a token, the token is addressed to its logical neighbour, irrespective of where it is physically located on the main cable. When the stations are first powered on, they will not be in the ring, so the Medium Access Control Protocol has provisions for adding and removing stations to and from the ring.

802.4 standards are very complex, much more so than 802.3, with 10 timers and some two dozen internal state variables. While 802.3 protocols are written in Pascal, 802.4 procedures are written in Ada. The token bus uses a 75 Ω broadband coaxial cable and both single- as well as dual-cable systems are allowed, with or without head-ends.

Token Ring (IEEE 802.5):

The token ring network architecture was introduced by IBM and IEEE has included the token ring as the 802.5 standard. A token ring is an interesting concept, in as much as it can be thought of as a point-to- point network, as well as a broadcast network. It is actually a collection of ring interfaces connected to each other, ultimately forming a ring. The network works in two modes, transmit and listen.

A particular station that wants to transmit captures a special bit pattern, called a token, that is circulating when the network is in an idle condition. When a station wants to transmit a frame, it is required to seize the token and remove it from the ring before transmitting. This is done by inverting a single bit in a 3-byte token, which instantly changes it into the first three bytes of a normal data frame.

Since there is only one token in the ring, at any instant only one station can transmit at any time. This totally eliminates the channel access problem. Naturally, all this implies that the ring must have a sufficient delay to contain a complete token to circulate when all stations are idle. To receive, which is done in the listen mode, the contents of the message are simply copied by the addressee.

Other salient facts about the token ring are the cabling can be twisted pair, coaxial cabling or fibre optic. The frame size is also not limited, because of its architecture. The current speed available on token ring networks is 16 Mbps. Its major disadvantage is that if there is any disruption in the ring, the entire network goes down. There have been, however, attempts to overcome this problem.

A major issue in the design and analysis of any ring network is the ‘physical length’ of a bit. If the data rate of the ring is R Mbps, a bit is despatched every 1 /R µsec. If the signal propagation rate is 200 m/sec. (a typical speed), each bit occupies 200/R metres on the ring. This will mean that a 1-Mbps ring whose circumference is 1000 metres can contain only 5 bits on it at one time.

A ring actually consists of a collection of ring interfaces connected by point-to-point lines. Each bit arriving at the interface is copied into a 1-bit buffer and then copied out onto the ring again. While in the buffer, the bit can be inspected and modified. This copying step introduces a 1 -bit delay at each interface.

Comparison between 802.3, 802.4 and 802.5 Standards:

To begin with the Ethernet standard, is most widely used, simple to implement and install. Where the network load is not very heavy and possibility of collision is low, it works very efficiently and with virtually no delay. Its major drawback is the fact that it uses large amounts of analogue components. For example, the entire collision detection mechanism is analogue.

Also, there are no priorities allotted to messages and an important message may get inordinately delayed due to this. It also has frame size limitations. No frame can be less than 64 bytes. As a result, even if the message consists of 1 byte, all 64 bytes get transmitted. This imposes a large and unnecessary overhead on the network and may also have some distance limitations.

A token bus (802.4) is highly reliable and easily installable. It can handle small frames and colli­sion avoidance is not required. It can priorities messages and thereby ensure that important messages are handled in time. Finally, it can support multiple channels. However, although the loss of tokens is infrequent, such losses create severe headaches for the system administrator.

Also the throughput, particularly at low loads, is bound to be low because transmission has to wait for the arrival of the to­ken; in general, the protocol is fairly complex. Also, it is not suited to fibre optics. As a result of these disadvantages, it is not widely used.

A token ring (802.5) is fairly easy to install because it uses point-to-point connections and it can be made totally digital. Because priorities can be used, the problem of delays of important transmission is not there. Finally, any transmission cable, including twisted pair, can be used (according to Tannenbaum, the choice varies from carrier pigeon to fibre optics).

Most of the important factors are summarised in Table 7.1:

While no final answer can be given, as to which of the three LANs is best, one can conclude that, as the wise man says, “it all depends.”

A conclusion drawn by Tannenbaum is worthy of note here:

“factors other than performance are probably more important while making a choice.”

FDDI:

802 LANs are based on copper wire (two copper wires for 802.6). These are invariably used for low speeds and short distances, but for high speeds and longer distances LANs must be based on fibre optics or highly parallel copper networks.

Fiber has the advantages that it provides high bandwidth, is lightweight and thin, is not affected by electromagnetic interference from heavy machinery power surges and lightning, and also provides adequate security, since it is almost impervious to wire tapping.

Naturally, therefore, fast LANs generally use fibre. Fibre Distributed Data some 1000 stations. It can be used in the same way as any of the 802 LANs. However, because of its extremely high speeds, a common use for it is as a backbone to connect copper LANs. FDDI II is the successor to FDDI I and has been modified to handle synchronous circuit-switched PCM data for ISDN traffic (or voice).

2. Computer Network # Metropolitan Area Networks (MAN):

In the development of networks, a Metropolitan Area Network (MAN) is one level above a LAN, in as much as it is spread over a geographical area, larger than a LAN. Typically, it is a network spread over an area the size of a large city and can occur, for example, if an organisation has several offices at several locations in a large city. In a MAN, usually data and voice are both integrated.

The origin of this arrangement is probably the use of computers by the police force of some large cities in the USA.

This was done in the earliest days of networking and the police found that in order to do their work efficiently, they needed voice as well as data communications between the various computers on the network. This also made economic sense, since the communication lines for a private network had invariably to be laid for MANs.

This meant that there was adequate provision on these communication lines for voice. The integration of voice with the data, therefore, came at no extra cost since the lines were laid out for data communications anyway.

The chief reason for separating MANs from LANs is a rather special way in which they handle their communications. MANs invariably use DQDB (Distributed Queue Dual Bus) technology for their communications.

This is displayed in Fig. 7.3:

DQDB is also known as the IEEE standard 802.6. In fact, it is the only standard defined by IEEE for MAN. In DQDB, two parallel buses run throughout the area (in which MAN is installed or to be installed). Each bus is unidirectional. Each computer in the network uses the top bus to send a message to a machine to its right and the bottom bus to send a message to its left.

Similarly, each machine uses the top bus (bus A) to receive messages from the machines to its left and the bottom bus (bus B) to receive messages from the machines to its right. Putting it more generally, we can say that each computer uses the bus in which it is ahead of the machine to which it wishes to send a message and uses the bus in which it is behind to receive the messages.

The head end of each bus generates cells in a continuous stream. The cell is 53 bytes in size of which 44 bytes contain data and 2 protocol bits—one indicating a busy status, that is, the cell has already been loaded and the second bit for a request.

The only information that the sender needs is to know’ whether to use bus A or bus B, depending on, on which bus is the addressee downstream. Data is loaded onto the packet and the protocol bits are set depending before the packet is inserted on to the bus.

Readers will observe that the transmission is a broadcast procedure. Automatically, the first machine wishing to send a communication becomes first in the queue and is served first. At the terminal point of each bus, the packet drops off.

There is also an interesting logic for queuing and servicing of each request in that the FIFO order is maintained in transmission without a central queue.

DQDB or IEEE 802.6, is probably the only manner in which MANs have been implemented throughout the world. WANs are widely utilised in the USA where they generally run up to about 100 miles (160 km) and T3 bandwidths (44.736 Mbps) (T3 is the same in the USA as E3 in India).

Computer Network # 3. Wide Area Networks (WAN):

Wide Area Network or WAN covers a very wide geographical area which could be across continents or oceans. The network consists of a certain number of hosts and the related client or user machines. Normally, a certain number of LANs are interconnected into a WAN. These hosts, usually there are several hosts in a Wide Area Network, are connected to each other through some communications medium.

The communications network is usually called a subnet. Usually, a Wide Area Network is designed in parts and the communications network design is separated from the design of the rest of the network. These are in any case two separate entities and treating them separately makes the design job much easier.

Usually, the communication lines are designed in steps and to send a message from one host to another, the communication lines (telephone lines, VSAT or any other means) are switched in between.

This switching between communication lines is handled by computers referred to as routers. This switching becomes necessary, particularly in Wide Area Networks, because invariably several trunks or channels are each terminated by a router.

Usually, the message is stored at each such router to be sent on the outgoing line later. This process was identified earlier as store-and-forward switching. In Wide Area Networks, unless satellite communications are being used, store-and-forward packet switching is usually used. Sometimes packets are small and fixed sized; they are then called cells. By their very nature, Wide Area Networks have irregular topologies.

Internets:

Internetworks, which we shall henceforth refer to as simply internets, which is a generic term and is different from the Internet, a specific internet that is spread worldwide. Internet refers to different networks that are temporarily connected to each other. Each network may be different from the other and, therefore, incompatible.

In order to connect two incompatible networks, a device called a gateway is used. The function of this gateway is to reconcile the incompatibilities within two incompatible networks. These incompatibilities may be in both, the hardware as well as the software. An internet, therefore, is a special kind of WAN. Subnets, consists of the communication lines and routers in a network.

If the hosts are added to these, then it becomes a network—a Wide Area Network, or, occasionally, a Local Area Network. On the other hand, if several LANs and/or WANs are connected through gateways then we have an internet.