19 Local Area Networks (LANs) We have seen how packet switching has contributed greatly to the efficiency and flexibility of ‘wide area’ data networks, involving a large number of devices spread at geographically diverse locations. Packet switching, however, is not so efficient for smaller scale networks, those limited to linking personal computers within an office building; that is the realm of an alternative type of packet-switched-like network called a local area network or LAN for short. In this chapter we discuss the concept of a LAN and the various technical realizations which are available. 19.1 THE EMERGENCE OF LANs LANs emerged in the late 1980s as the most important means of conveying data between different computers and computer peripheral devices (printer, file server, electronic mail server, fax gateway, host gateway, computer printer, scanner, etc.) within a single office, office building or small campus. LANs are constrained by their mode of operation to a geographically limited area, but are ideally suited for short distance data transport. A high bit speed LAN can carry high volumes of data with rapid response times. Such performance is crucial for most office applications, and has made them the ideal foundation for the new generation of ‘electronic offices’ comprising electronic work- stations, word processors, shared printers, electronic filing cabinets, electronic mail systems and so on. Most LANs conform to one of the different types specified in the Institution of Electrical and Electronic Engineers’ IEEE 802 series of standards. All the types have been developed from proprietary LANs, developed earlier by individual companies or organizations, but have now achieved American and worldwide recognition, as IS0 8802 standards. 19.2 LAN TOPOLOGIES AND STANDARDS The different types of LAN are characterized by their distinctive topologies. They all comprise a single transmission path interconnecting all the data terminal devices, with a 367 Networks and Telecommunications: Design and Operation, Second Edition. Martin P. Clark Copyright © 1991, 1997 John Wiley & Sons Ltd ISBNs: 0-471-97346-7 (Hardback); 0-470-84158-3 (Electronic) 368 LOCAL AREA NETWORKS (LANS) *c> (a) Star (b) Ring [c) Bus Figure 19.1 Alternative LAN topologies bit speed typically between 1 and 30 Mbit/s, together with appropriate protocols (called the logical link control and the medium access control (MAC)) to enable data transfer. The three most common topologies are illustrated in Figure 19.1, and are called the star, ring and bus topologies. Slightly different protocol standards apply to the different topologies. For example, IEEE 802.3 defines a physical layer protocol called CSMAjCD (carrier sense multiple access with collision detection) which may be used with a bus or star topology. Used with a bus form medium, such LANs are normally referred to as ethernets. IEEE 802.4 (IS0 8802.4) defines an alternative layer-l protocol for a token bus, again suitable for either a bus or star topology. IEEE 802.5 defines a layer 1 protocol suitable for use on a token ring topology. Finally, IEEE 802.2 (IS0 8802.2) defines a logical link control protocol (equivalent to the OS1 layer 2) that can be used with any of the above. This provides for the transfer of information between any two devices connected to the LAN. The information to be transported (i.e. information frame or packet) is submitted OS1 layer 2 Logical link control (IEEE 802.21 I CSMAlCD Token (IEEE 802.4 I i Token ring (IEEE 802.51 Physical network 'Bus'or'Star' I 'Token ring' I Figure 19.2 The IEEE 802 LAN standards CSMAjCD (IEEE 802.3, IS0 8802.3): ETHERNET 369 to the logical link control (LLC) layer together with the address of the device to which it is to be transmitted. Much like HDLC in X.25 (Chapter 18), the LLC assures successful transfer, error detection, retransmit, etc. Figure 19.2 shows the relationship of the various standards. Which physical layer protocol and which topology of LAN to use depend largely on individual preference and the compatibility of the existing computer kit needing to be connected to the LAN. To a lesser degree, the geographic circumstances and the network’s performance requirements are also factors. All the possible protocols transfer data between the nodes, using a packet mode of transmission; they differ in how they prevent more than one terminal using the bus or ring at the same time. The various protocols and their relative merits are now considered in turn. 19.3 CSMA/CD (IEEE 802.3, IS0 8802.3): ETHERNET CSMAICD stands for carrier sense multiple access with collision detection. It is a contention protocol. On a CSMA/CD LAN the terminals do not request permission from a central controller before transmitting data on the transmission channel; they contend for its use. Before transmitting a packet of data, a sending terminal ‘listens’ to check whether the path is already in use, and if so it waits before transmitting its data. Even when it starts to send data, it needs to continue checking the path to make sure that no other stations have started sending data at the same time. If the sending terminal’s output does not match that which it is simultaneously monitoring on the transmission path, it knows there has been a collision. To receive data, the medium access control (MAC) or layer 1 software in each terminal monitors the transmission path, decoding the destination address of each packet passing through to find out whether it is the intended destination. If it is, the data is read and decoded; if not, the data are ignored. The most important type of network that employs the CSMA/CD is called ethernet. Ethernet was originally a proprietary LAN standard (predating the IEEE 802.3 standard) developed by the Xerox corporation of USA. The original design was based on a length of coaxial cable, with ‘tee-offs’ to individual work stations, with a maximum of around 500 stations. The idea was to simplify the cabling needs of offices in which many personal computers were in use. Simply by laying a single coaxial cable along each of the corridors and connecting all the cables together, a bus could be created over which all the office computer devices could intercommunicate. Each time a new device was installed, a new tee-off could be installed from the corridor into the particular office where the device was situated (Figure 19.3(a)). Meanwhile, no new cabling needed to be installed along the corridor, so saving space in the conduits and averting the constant removal and replacement of the ceiling tiles. The technology for basic ethernet (lObase5) developed rapidly. First, clever devices for the tee-off points were developed, which enabled new devices to be connected very quickly without first severing the main coaxial cable bus. The devices pressed directly into the cable. This reduced the time needed for new installations and reduced the disturbance to existing users. Thin-ethernet (cheapernet or IObase2) appeared. This 370 LOCAL AREA NETWORKS (LANS) coaxial cable 0 - W- tee-off - baluns bus a) ethernet as coaxial cable bus b) ethernet as structured twisted pair cabling Figure 19.3 Typical coaxial cable and twisted pair wiring configurations for Ethernet allowed the use of narrower gauge coaxial cable as the main bus in smaller networks, and helped to reduce the installation costs. Meanwhile, the numbers of computer devices in offices were multiplying rapidly. Multiple ethernets became necessary, and increased flexibility was demanded to enable users to move offices without major cabling disturbances. This caused the development of LANs on structured cabling, using LAN hubs and twisted pair telephone cabling, in a star configuration. The ethernet ZObaseT standard was born (Figure 19.3(b)). As Figure 3(a) illustrates, in the coaxial cable realisation, a single cable bus, usually installed in the cable conduit in the office corridor provides the main network element. Tee-offs into individual offices are installed as needed, either by teeing directly into the main bus, or by using pre-installed sockets and connectors. A baluns, usually built into the coaxial cable socket in the end location, provides for correct impedance matching (50), whether or not the device is connected into the socket. When installed as part of a structured cabling scheme (nowadays the most common realization of ethernet), twisted pair cabling provides for the transmission medium. Multiple twisted pair cables are usually installed in each individual office and near each TOKEN BUS (IEEE 802.4, IS0 8802.4) 371 individual desk during office renovation, and wired back to a wiring cabinet, of which there is usually one per floor, installed in an equipment room. Usually next to the wir- ing cabinet, or even in the same rack, a LAN hub is installed. The hub replaces the coaxial cable backbone, so that the arrangement is sometimes referred to as a collapsed backbone topology. The bus topology still exists, but now only within the hub itself, which provides for the interconnection of all the devices forming the LAN, ensuring physical connection and appropriate electrical impedance matching. Should new devices need to be added, a spare cable can be patched through at the wiring cabinet and a new port card can be slid into the hub. Should any of the devices need to be moved from one office to another this can be achieved by re-patching at the wiring cabinet. The adds and changes are thus far less disruptive both to other LAN users and the office furnishings. In the structured cabling scheme, the baluns is no longer needed, since the hub provides for this function. Ethernet LAN components are relatively cheap. The bus topology is easy to realize and manage and is resilient to transmission line failures. As a result, ethernet has become the predominant type of LAN. The fact that any station may use the transmission path, so long as it was previously idle, means that fairly good use can be made of the LAN even when some destinations are unavailable because of a transmission path break, a capability which is not enjoyed by LANs employing more sophisticated data transmission, as we shall see later. Theory suggests that the random collisions of a large number of competing devices all trying to communicate over the same CDMA/CDLAN lead to rapid network performance degradation under heavy load. In practice, however, the traffic is rarely random, because most users communicate with the various main central server devices within the network which regulate the communication. However, should poor per- formance under heavy load be a problem, it can usually be overcome by subdivision into smaller, interconnected LANs. 19.4 TOKEN BUS (IEEE 802.4, IS0 8802.4) A token bus LAN controls the transmission of data onto the transmission path by the use of a single token. Only the terminal with the token may transmit packets onto the bus. The token can be made available to any terminal wishing to transmit data. When a terminal has the token it sends any data frames it has ready, and then passes the token on to the next terminal. To check that its successor has received the token correctly the terminal makes sure that the successor is transmitting data. If not, the successor is assumed to be on a failed part of the network, and to prevent ‘lock-up’ of the LAN, the original terminal creates a new successor by generating a new token. Transmission faults in the LAN bus can therefore be circumvented to some extent. However those parts of the LAN that are isolated from the token remain cut-off. Token bus networks are not commonly used in office environments where ethernet and token ring networks predominate. Token bus networks are most common in manufacturing premises, often operating as broadband (high speed) networks for the tooling and control of complex robotic machines. 372 LOCAL AREA NETWORKS (LANS) 19.5 TOKEN RING (IEEE 802.5) The token ring standard is similar in operation to the token bus, using the token to pass the ‘right to transmit data’ around each terminal on the ring in turn. The sequence of token passing is different: the token itself is used to carry the packet of data. The transmitting terminal sets the token’sflag, putting the destination address in the header to indicate that the token is full. The token is then passed around the ring from one terminal to the next. Each terminal checks whether the data is intended for it, and passes it on; sooner or later it reaches the destination terminal where the data is read. Receipt of the data is confirmed to the transmitter by changing a bit value in the token’s flag. When the token gets back to the transmitting terminal, the terminal is obliged to empty the token and pass it to the next terminal in the ring. The feature of IEEE 802.5 MAC protocol is its ability to establish priorities among the ring terminals. This it does through a set of priority indicators in the token. As the token is passed around the ring, any terminal may request its use on the next pass by putting a request of a given priority in the reservation field. Provided no other station makes a higher priority request, then access to the token is given next time around. The reservation field therefore gives a means of determining demand on the LAN at any moment by counting the number of requests in the flag, and in addition the system of prioritization ensures that terminals with the highest pre-assigned authority have the first turn. High speed operation of certain pre-determined, time-critical devices is likely to be crucial to the operation of the network as a whole, but they are unlikely to need the token on every pass, so that lower priority terminals have a chance to use the ring when the higher priority stations are not active. Token ring was developed by the IBM company, and is most common in office installations where large IBM mainframe and midrange computers (particularly AS400) are in use, in addition to large numbers of IBM PCs. The original form required specialized cabling (IBM type 1) and operated at 4 Mbit/s form. The initial idea was that a single cable loop could be laid through all the offices on a floor or in a building and devices added on demand. To avoid the disturbances and complications which might arise when connecting new devices to the ring (any break in the ring renders the LAN inoperative), IBM developed a sophisticated cabling system, including the various IBM special cables. The cable loop was pre-fitted with a number of sockets at all possible user device locations. The sockets ensured that when no device was connected, the ring was through-connected. However, on plugging in a new device, the ring is diverted through that device (Figure 19.4). The special socket for early token ring networks thus catered not only for correct impedance matching, but also for the ring continuity. Token ring cards in the individual end user computer devices connected to token ring LANs also need to be designed to ensure ring continuity in the case that the device is switched off. Thus the card reverts to a ‘switched-through’ state when no power is applied, so that even though the end device itself plays no active part in token passing while switched off, the tokens nonetheless still have a complete ring available. The further development of the token ring technology (mainly by IBM) has brought about the ability to use twistedpair cabling, and the emergence of a 16 Mbit/s as well as the original 4 Mbit/s version. In the 16 Mbit/s version, higher quality cabling (e.g. cate- gory 5 cable, as discussed in Chapter 8) may be required. TOKEN RING (IEEE 802.5) 373 0. unconnected Figure 19.4 Socket design in token ring LANs to ensure ring continuity Token ring LAN hubs have also developed alongside ethernet hubs, and allow for similar collapsed backbone topologies in conjunction with structured cabling systems. Thus a token ring LAN today is difficult to distinguish from an ethernet LAN (Figure 19.3(b)). The ring topology is collapsed into the hub itself, and two sets of wires to each individual user station allow for the extension of the ring to each user device. The switch-through function previously performed by the socket is also undertaken at the hub, so reducing the complexity and cost of individual sockets, so that standard telephone sockets may be used. The token ring LAN may differ from the ethernet LAN only in the port cards used within the hub and the LAN cards used in the individual PCs. Otherwise cabling, wiring cabinet and LAN hub unit may be identical. Indeed, in some companies, ethernet and token ring LANs exist alongside one another, without the user being aware to which type of LAN he is connected. Token rings, like ethernets, are common in office environments, linking personal computers for the purpose of data file transfer, electronic messaging, mainframe computer interaction or file sharing. Some LAN administrators are emotional about whether ethernet or token ring offers the best solution, but in reality for most office users there is little to choose between them. Token ring LANs perform better than ethernets at near full capacity or durihg overload but can be more difficult and costly to install, especially when only a small number of users are involved. 374 LOCAL AREA NETWORKS (LANS) In most cases, the choice between ethernet and token ring comes down to the recommendation of a user’s computer supplier, as hardware and software of a particular computer type may have been developed with one or other type of LAN in mind. Thus token ring remains the recommendation of the IBM company, whereas in all other environments ethernet has gained the upper hand. Slotted ring and other types of LAN also exist, but are not covered in detail in this book because they are rare. IS0 8802.7, for example, describes a slotted ring LAN used primarily by the UK academic community. 19.6 LOGICAL LINK CONTROL FOR LANs A local area network (LAN) provides for the establishment of direct (OS1 layer 2) connection between any two end devices directly connected to the LAN. However, although various different physical forms and topologies are possible (e.g. ethernet, token ring, etc.), it was quickly realized that all LANs are expected to be capable of the same basic function: carriage of data between software or applications running on two different computers. It therefore made sense to define a standard interface between the LAN and the computer software intended to communicate across it. This standard interface is called the logical link control (LLC) protocol. It is defined by IEEE standard 802.2 (IS0 8802.2). The logical link control (LLC) provides a standard communication interface equiva- lent to that provided by OS1 layer 2 to OS1 layer 3 (datalink service, see Chapter 9). LLC in combination with the medium access control (MAC) protocol specific to the particular LAN (e.g. ethernet, token bus, token ring) is equivalent to an OS1 layer 2 protocol. The information carried by LLC consists of four fields, which together are termed the LLC protocol data unit (PDU). The four fields are the destination service access point (DSAP), an address which identifies the application or software session to be activated in the destination computer to receive the packet the source service access point (SSAP), an address that identifies the application which sent the packet control information, which includes details of the type of connection (e.g. connection-mode, connectionless, acknowledged connectionless), the protocol in use at the next higher layer (e.g. TCP/IP, IPX, Appletalk, etc.) the user data (i.e. the raw data being transported) 19.7 LAN OPERATING SOFTWARE AND SERVERS So far we have talked about the physical structure of LANs, and the logical procedures used to convey the information packets across them. This alone, however, is not a sufficient basis for creation of an office LAN. In addition, a LAN operating system INTERCONNECTION OF LANS: BRIDGES, ROUTERS AND GATEWAYS 375 (software) is required. At the start, a number of different manufacturers offered altern- ative proprietary systems. Over time, the systems in use have reduced to five: Novell Netware, IBM LAN Manager, Appletalk, Windows for workgroups and WindowsNT. LAN operating systems provide for the software sockets (i.e. interface) between normal computer operating software (e.g. Microsoft DOS, Windows, Windows9.5, Apple Macintosh, etc.) and the new functions made possible by LAN networks (e.g. file server, host gateway, fax server, common printer, etc.). LAN operating systems are closely linked to network protocols, many of which have a proprietary nature. Thus, for example, Novell Netware (network operating system software) in conjunction with the Novell IPX network protocol allows the personal computer user to use various types of ‘network’ services. For example, a PC on the LAN may be able to choose between any of the printers connected to the LAN rather than being limited to the one directly connected to his computer. Sometimes he might choose the fast black and white laser printer, whereas on other occasions the colour printer is more appropriate. Alternatively, a common file server might allow the LAN users to share a common data filing system. In this way each individual user has a wider choice of facilities, and overall less equipment is needed because the printers and other devices can be shared between large groups of users. Equivalent functionality can be provided using the UNIX operating system software in conjunction with TCP/IP or the Apple Macintosh system in conjunction with Appletalk. Servers are typically powerful and expensive computers, capable of faster processing and additional functions useful to the workgroup as a whole. Servers are connected to the LAN, and usually remain in operation for 24 hours per day. Afile server is usually a computer with a large amount of storage capability which may be rapidly accessed and easily backed up by specialist computer staff on a once per day or once per week basis. It provides for secure storage of information and easy sharing of information basis on a workgroup or defined closed user group basis. A mail server provides for the transmission of electronic mail letters between individual PC users connected to the LAN without the need for the users both to be connected to the network and have their PCs switched on at the time of sending or receiving the letter. A facsimile server allows individual PC users to send printed documents directly to a remote facsimile machine without the need first to print the document to paper. The facsimile server itself is a device like a PC which is connected simultaneously to the LAN and to a facsimile/telephone connection. To the LAN, the facsimile server appears like a printer with LAN operating software, but instead of printing directly, the document is converted to facsimile (fax) format and transmitted over the telephone line to the remote fax device. 19.8 INTERCONNECTION OF LANs: BRIDGES, ROUTERS AND GATEWAYS The interconnection of numerous LANs, perhaps of different types, or the connection of a LAN to a mainframe computer or other external network or device requires the use of bridges, routers or gateways. We discuss these in turn. 376 LOCAL AREA NETWORKS (LANS) A bridge is used to link two separate LANs together as if they were a single LAN, typically enabling the maximum capacity of a single LAN to be surpassed, or two separate LANs in locations remote from one another to be connected as if they were a single LAN (a so-called remote bridge). The bridge is an intelligent hardware connected to the LAN, which examines the address in the LLC (logical link control) header of each packet or frame. For relevant frames, the packet is removed, passed across the bridge connection to the second bridge, where it is injected into the second LAN (Figure 19.5), which may have a different physical form (e.g. ethernetltoken ring). Either a table of the relevant addresses must be kept up to date in each of the bridges, to determine which packets must be transferred into the second LAN, or simply all packets are bridged. A remote bridge differs from a local bridge only in that a wide area type connection (e.g. X.25 connection or leased line connection) is used to connect the bridges. Usually only the packets destined for the remote LAN are bridged by a remote bridge, so that the lower bitrate of the bridge connection (typically 9600 bit/s) does not become a bottleneck. Although bridges provide for a relatively cheap means of interconnecting LANs, they are not to be recommended in large, complex networks, because they result in very complicated topologies which are extremely difficult to manage. Thus, for example, a bridge network of three LANs could have three bridge connections, connecting the LANs in a triangle fashion. The problem now is to ensure that the appropriate bridge connection (i.e. the direct one) is used when transporting frames between any pair of the LANs. In very large networks the chance of optimal path-finding is very low, so that there is a great risk of endless circular paths. To overcome this problem, the router appeared. Routers are much more intelligent devices than bridges. They are designed to ‘learn’ the topology of complicated networks (even ones which are constantly growing or changing) and accordingly route frames or packets across them to the destination indicated in the header. Routers learn about network changes through experience. Crudely put, if they receive a packet or message for a destination which they do not recognize, they choose a route at random and see if it is successful. On following occasions, the previously successful route is selected. In this way, communication is possible even across very complicated and cumbersome networks which have been built by different parties and simply connected together. Many LAN protocols (e.g. Novell’s ZPX, Appletalk, etc.) may be routed in their native (i.e. raw) form, but it is nowadays increasingly common instead to use the transmission control protocollinternet protocol (TCPIZP) as the main protocol to interconnect complicated LAN networks. / bridge Ethernet LAN Figure 19.5. LAN bridge