Chapter 10

Connectivity: Opening Up Your LAN


CONTENTS

A local area network (LAN) at its simplest is comprised of two or more computers connected with cables and network operating software (NOS), which allows all the computers to talk to each other through the cables. Maximizing computer resources and making files and applications easier to share are just some of the reasons for implementing a LAN. If you have two or 502 computer users creating documents and spreadsheets that they want to print, you don't need two or 502 printers. You can share as few or as many printers over the LAN as you choose. The same is true for the applications that the computers use. Instead of buying 502 word processing applications, you can buy one or two copies with the appropriate number of licenses for each user and share the single copy of the application over the network. Many businesses, government offices, schools, colleges, and even some home computer users utilize local area networks to get the most out of their available resources.

Selecting the best LAN for a particular environment is a lot like choosing a car. You must determine the basic functions you need and then work your way down to the details. For example, first you must decide what type of car to purchase. Once you determine that a Chevy Suburban will meet all your needs, you then begin to consider the important, but somewhat minor details, such as the color, type of interior fabric, and rear door options, right down to the type of radio you desire. The same is true for LANs, beginning with choosing the appropriate operating system and the cable connection, as well as choosing which services users may access and at what time they may do so.

Assume that you need to purchase printers for a LAN. Consider the following questions: How many users are on the network? How many printers are necessary for these users? Do any staffers need a personal printer to receive confidential figures? The type of network an office uses to share word processing documents across the room is not necessarily the best network to share medical CAT scan images across a hospital campus.

Determining your needs, the technology to meet those needs, and then setting up a LAN can be a daunting task. In many cases, it may seem as though you have too many options for some products and services, while in other cases, you may have only an option or two. Some background reading on today's trends and tomorrow's predicted trends is worthwhile before you begin laying the foundation for the LAN that will hold your intranet.

This chapter explores computer and network connections and the various hardware and software systems that are available for LANs. Additionally, I'll discuss the services these systems provide and how these services are useful for remote and telecommuter access.

Just as any good architect wouldn't design a "dream home" without getting to know her clients, you must first understand the needs of your LAN clients before you begin designing the LAN architecture. After fully accessing your users' needs, including their capabilities, the function of their work, and whether those functions might change in the future, you will be ready to start designing the LAN best suited for this group of clients.

LAN Hardware Components

Consider first what hardware comprises a LAN. Because the ultimate goal here is to connect all the computers via cables to form the LAN, exploring cabling options is a good place to start.

Connection Media

This section defines the physical cabling and connection techniques used to connect computers into a network. You will find at least one of these connection types in all LANs. In some cases, you will find multiple connection/cabling types. For instance, you may see fiber optic cables in one part of the LAN (perhaps a recently installed segment) and in another area find twisted pair cables. These two different media types cannot be directly connected. They are joined through a device that converts fiber optic signals to twisted pair signaling, such as a router. You'll read more about routers later. For now, let's look at the connection media in detail.

Coaxial Cable

Only a few types of cabling are used in LANs. We'll begin with coaxial cable. The first generation of local area networks utilized coaxial, or coax, cable to connect computers. Coax cable includes a center wire called a conductor, which is usually made of copper. The center conductor is then surrounded by an insulating foam or plastic covering. A foil or wire braid surrounds the insulation. The entire cable is then covered in a plastic sheath. The center conductor is the path for the electric signals (that is, data) on the network, while the foil or sheath is the ground for the electric signals. An example of coax cabling is cable TV, which generally uses this type of connection.

In fact, some local area networks are used for both cable TV and computer data. An advantage of coax cable is that the grounding foil or braid protects the electric signals on the cable from interference created by other electric sources. Electrical radiation from such sources as lights and air-conditioning units can corrupt the data signals as they travel down the cable. A disadvantage of coax is that the connectors can be difficult to install, and the cable itself can lose some of its transmitting quality if it gets squished or kinked, similar to the way a garden hose loses water pressure. Coax cable is still found in many network environments but is being replaced more often by twisted pair.

Twisted Pair

Twisted pair cabling is just that, a pair of wires each with a plastic coating that are twisted around each other and surrounded by an outside sheath. This is the same type of wire that brings phone service into your home. Twisted pair cables are easy to assemble and not as sensitive to kinks as coax cable. The main disadvantage of twisted pair cable is that it is more susceptible to electrical interference from other electric cables and systems and therefore has a lower transmission rate (that is, how fast the data can travel through the wire) than coax cable. However, to improve the transmission rates, a series of better insulated sheaths has been developed. These different categories of more efficient cables are referred to by their level of insulating properties. Category 1 cable, or Cat 1, has less insulating properties, while Cat 3 cable has better insulation; Cat 5 cable has even more insulation. These various levels of insulation for twisted pair cables allow for transmission rates comparable with coax cable, and in some cases, exceed the coax transmission rate.

Generally, the faster your LAN transmission rate is, the higher the category cable you will need.

Fiber Optics

A fiber optic cable is a very thin rod of flexible glass fiber surrounded by a plastic coating. Instead of electrical signals traveling down a wire as with coax or twisted pair cables, fiber optic cable uses light to transmit data. The most basic format of all computer "talk" is binary, a term represented in computer data as a 1 or a 0. Transmissions on a fiber optic cable take the form of either "light is on" to represent a 1 or "light is off" to represent a 0. In contrast, coax and twisted pair cable must represent ones and zeros with varying electrical voltages on the wire; for example, +5 volts for 1 and -5 volts for 0. Since the fiber optic transmission is light, and not an electrical voltage, it is subject to no electrical interference. This provides a much more reliable transmission rate even in electrically "noisy" environments. The downside to fiber is that it is expensive to implement, sometimes costing several times what coax or twisted pair cabling costs. The skill level required to fabricate the fiber optic LAN cable is also greater. For instance, the glass fiber must be cleaved and polished just right for the fiber optic connection to work properly. Twisted pair connectors on the other hand are simply fastened to the cable.

Wireless

Wireless LANs use a Radio Frequency (RF) transmitter and receiver at each computer instead of a cable. Each computer broadcasts and receives data to and from the other computers through the air much in the same way a radio works. This may seem like the easiest and most advanced way to connect computers to a LAN, especially with no unsightly wires and the ease of relocating workstations. Indeed many in the industry thought that wireless LANs would be the wave of the future. However, some issues must be considered with wireless technology.

The first consideration is that computers using wireless LAN technologies are broadcasting through the air to the other workstations on the LAN, as well as to anyone else who may be "listening." The data sent and received could easily be picked up and decoded by someone other than the recipient for whom it was intended. The reverse is also true. Your LAN may be subject to unwanted access attempts from someone who isn't even in the building or "connected" to the LAN. If doubtful, just listen to the numerous news reports of unauthorized persons acquiring cellular phone access codes who then use the codes to place thousands of dollars in phone calls. Cellular phones and wireless LANs both use RF signals, and more people are listening than you might expect.

Also consider that the airwaves are cluttered with TV, aircraft, police, CB, and many other radio signals. It is plausible that any of these RF sources could render your wireless LAN transmissions less than reliable or altogether impossible. Despite these drawbacks, there may be a place for wireless LAN applications-for users who face certain connectivity problems that would otherwise leave them unconnected.

These four connection technologies-coax, twisted pair, fiber optics, and wireless-make up virtually all the LAN connection techniques. The type of cabling used in a LAN is determined by such factors as environmental considerations, type of network operating system, transmission rate requirements, and even ease of maintenance. In the future, fiber optic cables will become more and more commonplace in local area networks. As the price of fiber optic connections decreases, they will replace twisted pair and coax cabling with increasing frequency, but coax and twisted pair will be with us for a long time to come.

There is some flexibility in terms of which cabling you can use in your LAN design. For instance, some LAN technologies may require twisted pair exclusively, while others can run over several different types of cabling. In the next section, we will look at the different types of networks that run over these cabling systems.

LAN Interfaces

Having discussed the basics of LAN cables, now I will explain how the cables actually connect to the computers. Typically the physical connection to the LAN is made through a Network Interface Card (NIC). This card also is sometimes called a Media Access Control (MAC) card. The MAC card is installed in the computer much the same as a video graphics adapter card or a CD-ROM controller card is. Some computers come with built-in network cards. A laptop computer can access a LAN with a PCMCIA LAN card or even through a parallel port adapter, which allows the computer to talk on the network through the computer's printer port.

Again, the functionality remains the same whether the computers use a parallel port adapter, a PCMCIA card, or a standard LAN card. The network card, in whatever form, provides a path for the data to travel from the computer to the LAN cabling. Four general types of network interface cards are used today in local area networks, and a few more new technologies are gaining popularity. The primary differences between them are in the way they "package" and transmit the data to the cable and the speed at which they transmit.

Ethernet/802.3

The Ethernet, or 802.3, network is probably the most commonly used network interface card. Ethernet networking was first developed by the DEC corporation. A variation on DEC's Ethernet technology was defined by the Institute of Electrical and Electronic Engineers (IEEE) in the IEEE 802.3 standard. The Ethernet card can transmit 10 million bits (a bit is a binary 1 or 0) per second on the LAN cable. Ethernet is sometimes called a 10 megabit per second (Mbps) LAN owing to this transmission rate. Recent advances in Ethernet technology have produced the 100Mbps Ethernet LAN, which supports the growing number of applications that need greater transmission rates.

Ethernet and 802.3 are sometimes used interchangeably, but there are differences in the way the two standards package the data, called frames, to be transmitted on to the network. In this discussion, I will use the term Ethernet to describe the 10Mbps network card whether it is sending DEC's Ethernet frames or 802.3 frames. Other frame types also can be transmitted with the Ethernet card, such as 802.2 (another IEEE standard) and SNAP II.

All of these frames can be transmitted at 10Mbps over an Ethernet card. The Ethernet network is a Carrier Sensing Multiple Access Collision Detection or CSMA/CD system. This means that the Ethernet cards in the computers constantly listen to the network cable to determine whether other computers are using the line. If a computer has data to send, the CSMA/CD system checks to see whether there are any current transmissions. If there is no activity on the line, the computer can transmit. If the line is busy, the computer waits until the line is free. If two computers talk at the same time, a collision warning is sent out on the network, and all the computers time out for a fraction of a second before trying to transmit again. The speed of the time-out and retransmit function is on the order of milliseconds, and typically, you will only experience delays of a second or so even in a very busy network. Ethernet is somewhat like an open discussion: If you want to talk, you need to wait until no one else is speaking.

Ethernet uses either a bus or a star topology. The bus topology is essentially a term that describes a single-cable LAN with all the computers connected to it. It is very easy to implement. The bus topology comes in two versions-thin coax cable and thick coax cable. Ethernet was originally implemented with thick coax cable, but the development of thin coax, also called "cheapernet," proved much easier to install because of its thinner, more flexible properties. The coax cable bus topology is a single cable that runs throughout the office or whatever area is being networked. You simply run the cable past each computer, add a connector to the cable, connect it to the computer and continue on with the cable to the next computer. The advantages of a bus cabling topology are that you only run one cable and that the coax cable used in the bus network is more resistant to electrical interference. The disadvantage of the bus network is also the single cable. If the single cable becomes damaged in any one place, all the workstations on the LAN will be affected.

The star topology uses twisted pair cable. In this topology, each computer has its own twisted pair cable that connects the Ethernet card to a connection port on a hub or concentrator, as shown in Figure 10.1. Inside the hub is the equivalent to the single coax cable which connects all the hub ports. Technically speaking, the twisted pair LAN is still a bus, but from the actual layout of the cable, it looks to be a star. The advantage of using a star topology is that if any one cable that runs to a computer is damaged, the other computers can still communicate over the LAN. The only real disadvantage is that you must run a separate cable for each workstation. Many LAN designers feel that the benefits of the twisted pair star far outweigh the disadvantages, and the twisted pair hub approach to Ethernetworking is very widely used.

Figure 10.1: An Ethernet network of the star topology type.

Token Ring

Token Ring was developed by IBM and is found mainly in mainframe or IBM AS/400 environments. Token Ring comes in two different versions: the 4Mbps and the 16Mbps version. As with Ethernet, Token Ring has an IEEE equivalent in the 802.5 standard, but they differ slightly in the way they package data into frames for transmission onto the network. The IBM Token Ring uses twisted pair cables and is set up in a star topology with each computer connected to a hub or concentrator. The 802.5 Token Ring also operates over twisted pair cables and can be in a star topology or an actual ring where each computer is connected to the next and the last computer is connected back to the first to close the ring, as shown in Figure 10.2. Unlike Ethernet, the communications over the Token Ring network are moderated by means of an electronic token-hence the name. Each computer on the network must wait to transmit data until it receives the token. Using this method, there is no possibility of data collisions created when two computers "talk" at the same time.

Figure 10.2: A Token Ring network.

ARCnet

ARCnet was developed in the late 1970s by the Datapoint Corporation. ARCnet uses a combination of Token Ring and Ethernet known as a token bus. The IEEE standard for token bus is 802.4. ARCnet transmits data at 2.5Mbps and can run over many different cables, including twisted pair and coax cables. The ARCnet technology is not used often in office networks but rather in factory and warehouse LANs. The ARCnet network is very forgiving in terms of the quality of the connectors used and in how far you can run the cables. Ethernet and Token Ring cables' limitations are measured in hundreds of feet as where ARCnet cable runs may exceed 20,000 feet. Due to the lower transmission speeds, many administrators faced with implementing a new network choose faster technologies that cost the same to install. In addition, many data product manufacturers do not provide ARCnet network equipment. For these reasons, I do not recommend ARCnet as a good solution for contemporary LANs.

FDDI

The Fiber Distributed Data Interface, or FDDI (pronounced "fiddy"), networking technology is a product of the American National Standards Institute. It is a 100Mbps LAN that uses a token-passing access scheme with a dual ring topology. The FDDI LAN has two fiber optic cables that connect to each workstation. This dual cable implementation provides redundancy in the event that one cable should fail or become disconnected. The FDDI network connection comes in two variations, as illustrated in Figure 10.3. The Dual Attached Station (DAS) connects to both fiber optic cables. The Single Attached Station (SAS) connects to only one of the FDDI cables.

Figure 10.3: A FDDI network.

Obviously, the SAS connection does not have the redundancy a dual station has. A FDDI LAN has many advantages. Because it includes no electrical signal in the cable, the data on a FDDI network is immune to interference from lights and other electrical sources. The data traveling down the fiber can't be "tapped" or monitored because the signal is light and doesn't radiate from the cable where it could be "picked up."

Indeed this security feature has sold FDDI into many government and commercial environments where security is a primary concern. The speed of transmission is also a major plus.Although with the new 100Mbps Ethernet technology, speed does not distinguish FDDI exclusively. FDDI also can be used as a Metropolitan Area Network (MAN), and some phone companies have begun marketing the FDDI technology to create what amounts to very large LANs for telephone customers. The main drawback of FDDI is the cost of implementing it over fiber optic cables. However, FDDI can be run over twisted pair cabling. In this environment it is called CDDI (Copper Distributed Data Interface).

ATM

Asynchronous Transfer Mode (ATM) is still largely uncharted territory because the technology is still in its infancy. The ATM implementation standards are still being defined. ATM is delivered over fiber optic cabling and has transmission speed capabilities in the 155 to 600Mbps range. ATM is a cell-switching transport service that packages data transmissions in small, fixed-size groups instead of variable-sized packets as in Ethernet or other LAN (and WAN) protocols. Using small, fixed-size cells over fiber optics allows for incredible transmission speeds. Some companies, including IBM, are rolling out ATM LAN applications, but for the most part, the cost of ATM is still prohibitive for most LAN applications. There are some ATM implementations in place supporting Wide Area Networking (WAN) and hospital or engineering labs that support medical imaging or computer-aided design applications. ATM is the future of LAN-based applications and services that require greater and greater transmission speeds.

Increasingly, more and more video teleconferencing, distance learning, voice communications, television, and graphic imagery will be running over local area networks, and we will need the transmission rates that ATM can provide. This kind of speed is crucial to supporting video and imagery data.

We have covered many different aspects of local area networking, including what kind of cabling and types of media access techniques can be employed to provide connectivity. As a rule, transmission rates of current LAN technologies are always being pushed to the limits as more and more speed-intensive applications find their way on to our computers. We will rely more and more on our computers for making voice calls, conducting videoconferences, and making bank and shopping transactions. The local area network will be tasked increasingly with these transactions.

In planing for the future, you must weigh the benefits of leading-edge technologies against the costs and decide whether you need the capabilities they bring. If you know that you will have to support computer design and modeling applications or video on the LAN in the next two years, you may want to look at ATM. Even though it will be more expensive initially, it may be cheaper in the long term versus installing an Ethernet only to dismantle it and upgrade to ATM later. We have discussed the prominent technologies, but by no means every type involved in local area networking. For the most part, these discussions cover a little of the past, the present, and the future of the local area network.

Any of these LAN networking techniques can be used in the building of an intranet. You must keep in mind, however, that you will be connecting LANs and that compatibility and end-to-end functionality is an issue. For instance, if you spend a tremendous amount of money installing 155Mbps ATM LANs in your offices to move huge imaging files around but can only afford a relatively slow T-1 (1.54Mbps) from the phone company to connect to your ATM LANs, you will have a bottleneck in your intranet, and the money you spent on ATM may be a waste. As with many things, your intranet may only be as fast as its weakest link.

LAN Software/Network Operating Systems

Now that we've covered how to connect all these computers and provided a way for them to talk to each other, we need to give them something to talk about. This section explores some of the different network operating systems (NOSs) that I mentioned earlier in this chapter. A network operating system performs the same function for the network that a Disk Operating System (DOS, UNIX, Mac, O/S, and so on) performs for a computer. The computer's operating system is the director of the computer's resources. It takes the data entered at the keyboard, displays it on the monitor, and then reads and writes files on the disk drives. When a file is printed, it feeds the data to the printer port. The computer operating system knows just where all of these resources are stored, how to identify the resources, and how to access the resources. Network operating systems perform the same functions but for many computers across a LAN. As mentioned previously, it's likely a group of users will share a printer, an application, or even a fax machine. To logistically ease the backup procedure, files generated by all the users usually are stored in one place on the network. The files necessary for the sales department to access are different from the files the accounting department needs, so certain users' access should be restricted, while setting permissions for other users will be necessary. Each of these procedures is directed through the network operating system.

The kinds of services you will find in a local area network, such as shared directories and printers or e-mail, varies from LAN to LAN. The primary services computer is called a file server, and all LANs have one. A file server is usually a computer much the same as all the other computers across the LAN. The file server, however, usually has a faster processor than the computers it serves, as it must "serve" the system's resources to all the computers. In addition to the basic file and printer sharing capabilities, a LAN can provide e-mail, group scheduling, World Wide Web access, and remote access. Providing these services to users often is as straightforward as loading software on the LAN.

Some of the more popular network operating systems are Novell, Banyan Vines, LANtastic, AppleTalk, and Windows NT. All of these operating systems provide the basic local area network services, such as sharing printers, directories, and files; they also allow access to other common resources, such as modems and faxes. In addition to the basic features, specific network operating systems often have advanced features or capabilities particular to that software. These more specific features may determine which NOS is utilized for each LAN. Some of these network operating systems are examined in more detail in the following paragraphs.

LAN operating systems basically fall into one of two groups. The first group is client/server. The client/server environment is distinguished by having one or more computers run software that provides the services for the LAN (for example, a file server or a print server). In the client/server group, LAN computers are loaded with software that allows them to access the server applications, hence the term client. The other class of LANs are called peer-to-peer networks. In a peer-to-peer network, any computer on the LAN can be serving applications or acting as the print server.

If you are planning to set up a LAN for 10 users so you can share files and printers, a 10Mbps Ethernet running a peer-to-peer operating system would be fine. If you are planning on getting 300 engineering, accounting, and management staff all connected to common resources, some with access to payroll or credit card accounts, you might consider a client/server operating system running on a FDDI network.

NetWare

NetWare is a network operating system designed by Novell, Inc. It actually is based on a networking technology developed by the Xerox Corporation called the Xerox Network System or XNS. NetWare first appeared in the early '80s with its early versions providing basicconnectivity for desktop personal computers, which were just becoming popular. The LANs of that time were small, usually consisting of all the same type of computers with little requirement or at least support for PC to Macintosh or UNIX system access. Typically, the office PCs needed to share files and printers, and interconnectivity between dissimilar computers or even security and auditing features were unheard of in LAN network operating systems. NetWare has undergone many revisions since then and is now a dominant force in the LAN industry, holding perhaps as much as 70 percent of the market share. NetWare is a client/server LAN system and can support Macintoshes.

The advantage of Novell is the capability to support many users and services. You could use Novell to connect different offices across the hall or across the country. Novell NetWare has the capability to support many different types of computers including DOS/Windows, Macintosh, and UNIX systems. NetWare also has options for accessing the local area network by dialing in with a modem and for setting up the TCP/IP protocol for access to the Internet. It is ideal for mid-size to large size LANs and LANs with all different types of computers. The disadvantage of NetWare is that it takes a while for both the LAN administrators and users to become proficient in using all this functionality. NetWare can be run over Token Ring, Ethernet, ARCnet, and FDDI LANs.

Windows NT

Microsoft's Windows NT is the latest LAN operating system to appear on the market. Windows NT is a client/server operating system designed to support LAN connectivity in much the same fashion as NetWare. Windows NT has many of the same features that NetWare has and is NetWare's biggest competitor. Windows NT supports dial-in access to the LAN and Internet access support. Advantages of Windows NT are that it is fairly easy to become proficient in, and in many cases, it is less expensive. It doesn't have the long-term track record that many of the other operating systems have, but it is enjoying a very enthusiastic and growing loyal customer base. Windows NT can also be run over Token Ring, Ethernet, ARCnet, and FDDI LANs.

Windows for Workgroups

Windows for Workgroups is Microsoft's peer-to-peer networking environment. This NOS is good for small LANs that need to provide very basic connectivity and services, such as shared directories, printing, and application serving. Windows for Workgroups is very inexpensive compared to client/server LAN packages but again does not provide specifically for the more advanced features, such as security auditing and tracking that the larger client/server LANs offer. Windows for Workgroups runs over Token Ring, Ethernet, ARCnet, or FDDI LANs.

LANtastic

LANtastic is a peer-to-peer LAN that actually uses some of the NetWare protocols to communicate over the network. This is another NOS that is good for small LANs. Basic connectivity and services such as shared directories, printing, and application serving are supported. LANtastic is inexpensive and easy to set up compared to client/server LAN packages but again, does not provide as many service options as the client/server LANs, or even as many as Windows for Workgroups. LANtastic runs over Token Ring, Ethernet, ARCnet, and FDDI LANs.

AppleTalk

AppleTalk is a peer-to-peer LAN that is specific to Macintosh computers. AppleTalk is good for small Macintosh LANs. It supports basic connectivity and services such as shared directories, printing, and application serving. AppleTalk has no support for systems other than Macintosh. It is inexpensive because it is part of the Macintosh computer operating system and is easy to implement.

TCP/IP

TCP/IP is the "language" of the Internet. It is a networking technology developed by the United States Government Defense Advanced Research Project Agency (DARPA) in the 1970s. It is most commonly employed to provide access to the Internet but can be and is used by many people to create a LAN that may or may not connect to the Internet. In many aspects TCP/IP is a client/server-type LAN, but many manufacturers of TCP/IP software have applications that allow the "clients" to serve files or even applications.

TCP/IP has, from an architectural standpoint, some of the same qualities as NetWare, but TCP/IP is truly an open systems protocol. This means that no one manufacturer creates the product-any computer running TCP/IP software can connect to anyone else who has TCP/IP software (provided the user has an account and security permissions), regardless of who made the particular version of software. For instance, the FTP Software company's version of TCP/IP, called PC/TCP, is completely compatible with the Wollengong company's version. Novell also makes a version of TCP/IP software to provide Internet access functionality for their client/server NetWare LAN, and a version of TCP/IP is completely compatible with the PC/TCP software from the FTP company. TCP/IP is an ideal intranetworking protocol. If you have two different type of LANs such as Windows NT and NetWare which do not "talk" to each other, you could run TCP/IP in addition to the NT and NetWare software to get inter-operability. TCP/IP runs over all networks, including Token Ring, Ethernet, ARCnet, and FDDI LANs.

These are the more popular LANs in use today. Support for these networks is widespread with many publications, Internet discussion groups, and certification programs for those who install and maintain these networks. Novell and Microsoft have established Certified Network Administrator and Certified Network Engineer programs.

Other LAN operating systems include

None of these other LAN systems have had the impact that NetWare, TCP/IP, and Windows NT have had on local area networking. There are certainly large numbers of DECnet or Banyan Vines LANs being used today, but many consider the serious client/server LAN to be a "neck-and-neck" contest between NetWare and Windows NT.

Remote and Telecommuting Access to Your LAN

I have described some of the services found in the local area network, such as the file serving, print sharing, or e-mail in brief, and I have put these services into the context of a local environment, such as an office. One of the services mentioned is the remote access of the LAN resources-in other words, dialing into the LAN with a modem from a remote location to share files or perform administrative tasks. The idea of remote access has long been a part of LAN services, but with recent advances in the speed of modems, new protocols such as PPP (Point-to-Point Protocol), and phone company services, such as Integrated Services Digital Networking or ISDN, remote access functionality has improved dramatically. Just a few years ago, a remote user would dial into a computer on the LAN at 2400bps and log into the file server from the computer on the LAN. The remote user could only see text displayed through this kind of link, and the types of transactions performed were limited because the connection was very slow. Today, a remote user can log into the LAN and perform virtually all types of transactions as if he were actually in the office. In this section, we will discuss these different types of connections and how the LAN can be expanded or opened to provide remote access service.

Modems

Modems have been used in computer communications for many years now. A modem is a modulator/demodulator. It takes the binary ones and zeros that all computers "speak" and turns them into a series of audible tones that can be sent over standard phone lines. (Standard phone lines are also referred to as plain old telephone service or POTS lines.) Modems are phones for computers, and they basically work in pairs just as phones do: the modem you are using to call and the modem that is answering your call. You can connect to any computer in the world that has a modem, just as you can call any house in the world that has a phone.

A modem is connected to a computer's communications port, known as a comm port (such as COM1 or COM2), and also plugs into a telephone outlet the same way a phone does. Modems are also available as a card that plugs into the computer in a manner similar to a LAN card. The computer is loaded with software, such as Datastorm's Procomm or PC Anywhere, which operates the modem and allows you to dial up the modem (that is, computer) with which you wish to communicate. Modems modulate and demodulate the data bits (1s and 0s) very slowly. If you recall, the Ethernet transmission rate is 10 million bits per second. The maximum modem transmission rate over phone lines is 28,800bps. This limits the type of transactions that can be performed over a modem, particularly in terms of accessing LAN services that are designed to be served at much faster speeds.

If you are retrieving a business letter from the file server to your home computer with a modem, the transmission rate is probably acceptable. If, on the other hand, you are trying to use the LAN word processor application on the file server through a modem connection, the transmission rate will be slow enough to make the process a very tedious venture. Modems are very inefficient for computer transmissions. All the modulating and demodulating takes time to perform; this is one of the reasons modems might never be able to exceed 28.8Kbps.

Traditionally, remote users have used a modem to access the on-site LAN by calling one of the LAN's computers, which has modem for such a purpose. This computer is called a communications server or gateway, and it gives the remote user access to the LAN by using the comm server's LAN software and network connection. The comm server is really just another LAN workstation, except it has a modem and modem software on it. One of the disadvantages of this type of access is that you must first load any files you wish to send to the file server on to the comm server through the modem connection and then from the comm server to the file server.

The file server doesn't recognize your home computer through a modem connection as a LAN client. You must use the comm server to act as a LAN client on your behalf. You could set up the file server as the comm server to gain direct access to applications and files, but this bypasses the client-to-server connection process (called a login) and poses definite security risks. You may not have network security concerns per se, but it is still a good idea from a network administrator's point of view to have a dedicated comm server.

As you may already have imagined, it can become a bit of a headache setting up a comm server that can support dozens or even hundreds of dial-in users. The idea of one LAN workstation with a few modems attached to it is functional for a very small LAN, but would be nothing short of a nightmare for a LAN that supports 120 dial-in users.

Another approach to dialing in with modems is the modem bank and terminal server. In this scenario, the comm server still uses modems, although they are modular or rack mountable in design versus a desktop or internal PC type modem. This allows you to add new modems in an orderly fashion as the demand grows. The modular modems are mounted in an equipment rack where each is connected to a phone line and to the communications server, which for this type of connection is actually a device called a terminal server. The terminal server is not a computer per se as in the earlier scenario. The terminal server does provide access for the incoming call to the LAN, but it also adds the ability to negotiate a more sophisticated type of connection.

In the modem bank/terminal server type of connection, the computer dialing in can be assigned an actual LAN identity or address. This allows the computer to converse directly with the LAN servers and resources. This method of connecting is called remote client access because the dialed-in computer becomes a LAN client as if it were actually on the LAN. This type of connection is negotiated using a protocol called PPP or Point-to-Point Protocol. Both the terminal server and the user's home computer must have the PPP software to establish this connection. Although the speed of the connection is still limited to the transmission rate of the modem, PPP is still a more "graceful" connection and is far more streamlined than having a comm server act as a "middleman" in the connection.

ISDN

You may have heard of the phone company's ISDN, or Integrated Services Digital Network. This type of networking provides a digital or binary (all 1s and 0s) transmission service that can be dialed just as a standard phone (POTS) line can be dialed. The advantages of ISDN for LAN access are extensive: Your computer doesn't have to have its binary "talk" modulated into tones for transmission over phone lines as modems do. The data stays digital from your computer through the phone lines to your LAN. The slowest ISDN speed is 64,000 bps or 64Kbps. The higher ISDN speeds exceed one million bps. ISDN generally comes in two forms. The first is the Basic Rate Interface or BRI. BRI is two 64Kbps channels called bearer or B channels. Both these B channels can be used together to form a 128Kbps connection. The second type of ISDN service is the Primary Rate Interface or PRI. A PRI is a 1.5 million bps line (1.5Mbps) that is made up of 23 B channels.

ISDN requires different equipment than POTS lines. A modem will not work with an ISDN circuit, because the ISDN service is all digital, whereas the modem is designed to turn digital "talk" into sounds or analog "talk." The type of equipment used to access ISDN depends on whether it is a BRI or PRI service. For the BRI service, you need a terminal adapter or TA. A TA connects your computer to the ISDN line. As with modems, you also need software to dial over the ISDN service. PRI service, on the other hand, is a more sophisticated and complicated kind of service that requires something called a channel service unit, or CSU, to help sort out the 23 channels of data that make up the PRI service.

The ISDN access technique at it simplest is much the same as it would be with a modem connection. One ISDN-connected computer dials another ISDN-connected computer-only the speeds involved are much faster. You can buy an ISDN card that installs in the computers or an ISDN TA that sits on the desk just as a modem does. This is a typical implementation for simple connectivity between computers. Using ISDN, you would dial directly to the file server, which would also be equipped with an ISDN card to access services in this way, but the number of incoming calls to the LAN would be limited by the number of ISDN cards in your server.

A more secure and functional method for connection to the LAN would be the ISDN terminal server solution. This solution also uses the Point-to-Point Protocol, but the phone line is the higher speed ISDN. An ISDN terminal server with PPP provides dial-in users with the feel of actually being directly connected to the LAN. The response time in accessing LAN resources is greatly improved with this solution.

The use of ISDN is expanding every day. In previous years, only very expensive digital services were available from the phone companies, and companies leased the digital line on a full-time (24 hours a day) basis whether they used the line or not. The leased 56Kbps is an example of the full-time digital line. ISDN has brought users the "dial on demand" services that allow them to dial the connection when they want it and pay per use. Even the most basic ISDN service (BRI or 128K circuit) provides a better transmission rate than the slower, more expensive 56Kbps service.

Setting Up the Network Servers

There are many things to consider in setting up the network. After you have determined the number of users and what type of LAN technologies you will use, you must select the types of hardware (that is, which computers, cables, and network cards). A fair amount of these decisions can be based on cost, but you must also consider the reliability of the equipment. If your file server keeps breaking down, it is hard to justify having saved $300 on the purchase when you've spent twice that much time fixing it. Furthermore, as soon as your users are acclimated to the LAN and using its services, even the shortest downtime will generate tons of phone calls to your office. Spending a little more on better equipment helps reduce problems in the long run.

You must also decide how to route the cabling through your facility. You could simply lay the cables around the floor, but this isn't pleasant to look at and will probably cause problems with people tripping over the cables and disconnecting them. You could place the cables up in the ceiling, but that could also cause problems because this puts the network cables close to electrical cables, possibly causing data transmission problems. If a lot of network cables are necessary to connect all your computers, you may not be able to trace or isolate a cable problem if the cables are jumbled together.

Some advanced planning is in order for the layout of the cabling. You may want to have a structured cabling system installed. A structured cabling system is a generic cabling system that can be used to connect data video and voice lines throughout your facility. It is a little more expensive for a LAN implementation only, but if in the near future you'll need to install a new phone system to support your growing business, you may be able to kill two birds with one stone and install a cabling system that supports both.

After you have decided what your users need to do and what LAN systems support those requirements, you may want to have an independent consultant look at the design to confirm your plan. A few hours time spent with someone knowledgeable about LANs and the applications that operate over them may be well worth the money. At the very least, you can confirm your LAN design skills, and if your design is incomplete, you have the opportunity to gain some valuable information.

After you are satisfied that your LAN design is adequate, you must evaluate the three issues surrounding the actual implementation of the LAN. The first is determining whether you really have the resources "in house" to perform the installation of all the network cards and cables. The second is setting up the software on the servers and client workstations and getting them to work together. The final issue is the administration of the network. Are you prepared to handle all the trouble calls and requests for additional services?

The server must have routine maintenance check-ups, and data files must be backed up periodically in the event of a system failure. All this day-to-day operations "stuff" is a full-time job and then some. You may very well want to contract out the installation, integration, the administration, and even the LAN design to someone with experience in networking. Many organizations, both commercial and in government, do just that.

You could have the administration portion of the network staffed by contractors until you have hired a few people into your staff to take over. When selecting contractors for the installation and administration of your network, make sure they have experience with the network you are installing. The same applies for those you hire as LAN support technicians. They should be able to answer detailed questions and demonstrate a clear understanding of the LAN systems you have implemented. Never allow one person to become the sole caretaker, administrator, or technician for your network. If that one person leaves your organization or if that person is away on vacation and the network goes down, you may find yourself with a network that doesn't work, and no one who knows what the cause might be.

At a minimum, your network will have one server (even in a peer-to-peer network), which will be hosting the applications or files. More often than not, a LAN has several servers to provide all of the services your users will require. You will more than likely be using an Intel-based computer (this would be the kind of computer with the Intel processor chip in it) and running DOS and Windows. The specific LAN software you select will have specifications for the minimum system requirements, and it is recommended that you exceed them where possible. If the LAN package requires a 486 66MHz computer with 8MB of RAM, you should probably go ahead and get a Pentium 90 with 16MB RAM, just to be on the safe side.

Network servers can be located anywhere on the network; however, the servers should be located off the beaten path. That is to say, you shouldn't have the servers located where people are going to be bumping into them or setting their morning coffee on top of them. These machines are the most important pieces of your LAN. You'll want to purchase a good uninterruptible power supply (UPS) to protect your server against power surges or outages. Locate the servers in a cool, environmentally controlled room. You may need air conditioning if the room has a great deal of equipment and is warm or not well-ventilated. In large LANs, rooms with many servers are referred to as server farms, and they often have not only air conditioning systems, but also humidity control and fire control systems.

Servers need to have their data backed up regularly. This typically involves a cassette tape-like system that is installed in the server, similar to a floppy or CD-ROM drive. A copy of the server files are made with the tape backup every day or every other day depending on the amount of data being created. If you are selling widgets by the thousands every day and the accounting for the business is on the LAN, you want to back up the accounts payable and receivable, as well as the inventory database every day. Then once a month, back up the entire server, applications and all. In fact, make two copies of your backups. Store one in the LAN administrator's office and the other off site. In the event of a fire or other disaster, having a backup at your house will save a lot of suffering. Some administrators deposit monthly server backups in safe- deposit boxes at two different banks in addition to the two office copies. If you are serious about your LAN, take your backups seriously.

The next few sections contain some information on the different types of servers you will find in a LAN. They are not necessarily different computers. The file server may also be the print server or the e-mail server. The different servers are listed by the services they perform to help define the roll they play in the LAN.

File Servers

The file server is the heart of the LAN. It has the applications that the users need and the files that are shared. It is usually divided into sections that the users can access, such as their own directories or shared directories, and the sections that are restricted to administrative access, such as security directories and operating system files. The computers with client software can view a list of servers on the network and select the one to which they will connect. Often each department, such as engineering, accounting, or sales, has its own server. Users may or may not have connection or login privileges on every server. Someone in engineering may have no reason to have an account on the accounting server. The servers also need to be serviced periodically. Running a disk optimizer program or utility to check the health of the disk is recommended. Norton Utilities is an example of a system diagnostic/preventive maintenance package that is widely used on DOS-based computers. Many network operating systems have their own disk maintenance programs.

Print Servers

Print servers are the servers that have a printer or printers attached to them. In a LAN environment, a computer prints to a network printer over the LAN cabling. You could make the file server also the print server, but you'll want to locate the file servers some place out of the way. You'll need to have a server that will satisfy the printing requirements located near the users. A print server is just that. The software in the client computers redirects the print jobs that would normally go to the LPT port out on the user's computer out to the network print server.

If your network is small-say, 20 users-you probably just need one print server with one printer. If your network is sixty people on two different floors, all printing documents and some printing color slide presentations, you probably want three or four print servers with five or six printers. A print server accepts the print jobs in the order in which they are sent and stores the files. This is called spooling a job. As soon as the printer can accept a new job, the file is sent out the print server's LPT port to be printed.

E-Mail Servers

E-mail is quickly becoming the way businesses communicate. It has in many cases replaced voice mail. E-mail comes in two basic formats: the version that is designed to work within your LAN using your LAN's software protocols (such as Novell's Groupwise Messaging System), and a TCP/IP based system that allows Internet e-mail as well as internal LAN mail. The Simple Mail Transfer Protocol, or SMTP, is an example of the TCP/IP e-mail service. The e-mail found on the Internet is largely SMTP mail. Many LAN operating systems such as Novell's Groupwise Messaging have add-on packages that allow access to Internet e-mail by creating gateway services. A gateway takes one kind of application, such as the Groupwise Message e-mail, and converts it into the SMTP e-mail format so it can be sent out to the Internet. The e-mail server can be located on the file server, consolidating the backup process.

Communications Server

We've talked about users dialing into the network to access services. The communications server is the one that provides that functionality. Again, this type of server for dial-in services is only really practical for the smaller or mid-size LANs. Larger LANs may require a modem bank and terminal server for the more sophisticated access discussed earlier. However, a comm server can provide a very useful service in addition to dial-in support-namely, dial-out. You may want to fax documents right from the desktop across the LAN through the comm server. With a fax modem attached to the server and some fax software, your users never need to walk over to the fax machine. The reverse is also true. A user can receive faxes into his or her own mailbox through the comm server. Support for this type of functionality varies with the different LAN operating systems, so you may want to make this an element of your LAN software selection process.

The communications server can also be used to perform EDI transactions. EDI, or Electronic Data Interexchange, is an electronic commerce application that allows organizations to share business information via computer systems rather than mailing large documents, such as proposals and contracts, through the postal system. You simply post the information on your comm server and other computers can dial in to retrieve a copy. The reverse application would be configuring your comm server to automatically dial into another organization's EDI comm server to fetch the latest documents. EDI is an effective way to gather or disseminate information to large numbers of people with minimum effort.

World Wide Web Servers

Many organizations are using the Internet World Wide Web (WWW) service to provide information about their organization's services and products. The WWW format displays electronic documents or pages with graphics that can be accessed using the TCP/IP protocol. The WWW service uses a protocol called Hypertext Transfer Protocol or HTTP. This page layout is useful in displaying pictures of products, places, or people, and the accompanying text can be set up to link one page or Web site to another page or site. As you view the page, the link text is highlighted so that when you click on the text with your mouse, you jump to the other page or Web site automatically.

Typically, a Web server is accessed across the Internet, but it can be used within a LAN without Internet connectivity. You still must have TCP/IP software running on the workstations. A Web server can be set up on your file server, but if the Web site becomes extremely busy (particularly if it is connected to the Internet), the serving of files and other services to your users may slow down.

Network Connectivity Hardware

There is some additional hardware that you can use in local area networking, some of which I mentioned earlier. This includes the hubs that are used in the twisted pair Ethernet LAN. The hub is the device to which all the twisted pair LAN computers are connected. These hubs come in a variety of makes and models, and selecting one can be based largely on price. However, as with the servers, you probably don't want the cheapest one in your LAN. A good rule is to eliminate the cheapest and the most expensive models and what's left will probably serve the average LAN pretty well.

A variation on the hub concept is the switch. A switch has ports just as the hub does, but the switch is smarter than a hub. The switch learns all the addresses (the unique network identifying numbers that are on each network interface card) of all the computers on the LAN. The switch knows what address can be found on each of its ports. When you issue a login or connect request to a server, the switch sees the request and connects the calling port directly to the called port. This is similar to the phone company's switching. The advantage is that theconnect request doesn't have to circulate around the network looking for the right address. The switch knows which port the server is on and connects you directly.

Because the switch is handling all the connections, fewer transmission problems occur, and more time is spent sending data instead of collision messages. Switches were first used for Ethernet networks, but now there are switches for Token Ring LANs as well. If you are supporting the heavy-duty applications, such as Computer Aided Design and Software Engineering packages on your LAN, or if your LAN is simply very busy with several users, you may be a candidate for a switch. If users are complaining that it takes forever to send and receive data and that the collision lights on your equipment are constantly flickering, you may need a switched LAN.

There is a lot of leeway in the day-to-day operations of the local area network. Every little connector and cable need not be the same type or engineered down to the micrometer. Every workstation need not be set up exactly the same in terms of operating system and hardware performance. Establishing solid maintenance schedules and backup routines will go a long way in keeping things running smooth.

Summary

You've just gotten through an overview of one of the more intimidating aspects of intranets-the network hardware. This is a subject that whole books are written on (try Understanding Local Area Networks or Understanding Data Communications, 5th Edition, both from Sams Publishing), so you have some more learning to do.

But by now you should have more than a little knowledge of LANs. In the next chapter, you'll expand upon this knowledge and learn the principles of taking your LAN out into the big, wide world to meet other LANs-forming a wide area network.