Small businesses generally lack support from a corporate IT office. The ideal network for such an organization therefore must be straightforward and easily managed. No two organizations have the same needs, and complications will always arise. One of the guiding principles throughout this chapter will be to follow the simplest approach to achieve the desired results.
No matter the size of an organization, it must address the provision of computer support. A small business may have only one person dealing with computer issues-and then often on only a part-time basis. Many organizations hire a consultant to establish their computer environment, and then rely on in-house staff to keep it going.
Small Office Computing vs. Enterprise Computing
Small office computing has a character quite different from the computing environments that support large organizations, often called enterprise networks. Enterprise networks may have thousands of users, and involve a complex array of servers, mainframe systems, wide-area network links and the like. An enterprise network may serve multiple geographical locations and multiple buildings at each location. It is not unusual for an enterprise network to include several thousand devices. Such a network relies on a backbone network that channels data among locations and local area networks at each site. An enterprise network includes sophisticated equipment that must be maintained by highly trained network administrators.
Smaller organizations have more modest computing and networking requirements. They might have a dozen or so computers and a few laser printers. The network for the small office must allow members of the organization to share information, as well as printers and other peripherals. The computing needs of most small organizations can be met by a single LAN with one or two servers, using off-the-shelf components. Unlike the enterprise network, a small office LAN usually can be managed by one person with only moderate technical knowledge and experience.
While the small office network doesn't match the scale of its enterprise cousin, many of the same issues apply to both. The design of a small network must be simple, yet functional, secure and scalable. As the business grows, the network must easily expand with it. Even if the scale of the initial environment is small, avoid making technology decisions that might limit your company as it expands.
Assessing Functional Requirements
Before you begin designing your network, have a clear sense of what you need it to accomplish. One prerequisite to network design is a complete assessment of expected functionality. Set aside some time to think about all the tasks you want to automate or make more efficient through your computer network. What business applications do you need to support? Do you simply need to provide shared access to word processing files, or do you have multiuser databases to support? Do you need electronic mail? Web servers? Point-of-sale operations? Will you require Internet access? Once you have considered all the business tasks and functions you expect to implement on the network, write them down and assign priorities to each item. As you begin deploying your plan, you might need to consider which parts you can do now and which can be addressed later. Take care of critical business functions first.
Sizing the Network
Have a clear idea of your network's expected size, considering its number of users and their intensity of use. Be sure to plan for future growth by building in lots of extra capacity from the beginning. Calculate what capacity you might need in two or three years. Consider the number of new users as well as dramatic increases in data storage needs per user. Your network should be designed to grow easily with incremental additions of existing technologies.
Follow a Standard Approach
It is important that you build your small business network using standard, industry-proven components. As business relationships change, you may need to interconnect your network with others. Protect your investment by building a network that is likely not to pose compatibility problems. If you are an independent branch of a larger organization, be consistent with umbrella group's practices and standards. Even if you are expected to maintain a separate network today, you may need to be part of its wide-area network in the future.
What types of external connections will your network need? Is Internet access necessary? If so, will a dial-up connection suffice, or will you need a full-time dedicated link? How much bandwidth? Will you need to connect with private networks, such as your home office network? One of the most challenging aspects of the small office network involves setting up links to external networks. Not only are these the most technically complex tasks of implementing a network, but they also carry significant costs.
Creating the General Design
Once you have assessed the new network's functional requirements and relative scale, you are ready to begin the design work. Network design involves several layers. You will need to make decisions on each of the following:
- Network type. Options include Ethernet, ATM or token ring. Most small networks are based on Ethernet, but even within this category there are options: Shared media 10Base-T, switched 10Base-T, shared media 100Base-T, and switched 100Base-T. To make a decision, you will need to scrutinize the relative bandwidth the network must support. The greater your need to support multimedia applications such as streaming audio or video, the more you will need a pricier high-bandwidth solution.
- The physical network. This includes network cabling, faceplates, and other issues of basic infrastructure. The kind of cabling you install depends on the network type you selected.
- Network communications equipment. To operate the network, you will need devices such as Ethernet hubs and routers.
- Network operating system. Currently, Microsoft Windows NT Server and Novell NetWare dominate this area. Some environments may require Unix-based servers. It is also possible to design a peer-to-peer network based on NT Workstation.
- Network server hardware.
- Data backup hardware and software.
- Client workstations. Consider the hardware (PC, Mac, etc.) and operating system (Windows 95 or 98, Windows NT, MacOS, etc.).
Final Design: Making Technology Choices
In the early design phase, we were painting in broad strokes. Now we must consider each aspect in detail. Each section below will take a closer look at the technology choices available, and describe how they apply to the small business network environment.
One of the first decisions in computer environment design is the selection of the network type- a group of products that work together, even if they are manufactured by different companies. Products in the same group each follow the same networking rules, and you can count on them to work together properly. Today's most common network types include Ethernet, token ring, and ATM. Each of these three offers a viable alternative for supporting a LAN, each with its own costs and performance benefits. As we will see, Ethernet stands as the prevailing technology and generally is the most appropriate choice for small business networks. Once you select a network type, the network cards, cabling and network software you choose must be compatible with that group.
Token-ring networks can be found primarily in environments with a significant amount of IBM equipment. This network type uses a token-passing protocol; each computer communicates on the network only when presented with the network token. Computers read incoming data packets and transmit outgoing ones as the token rotates throughout the network. Token-ring networks became popular in organizations using IBM mainframes, and continue to be used to a limited degree. At one time, token-ring networks outperformed Ethernet, but this is no longer the case. Token-ring network cards are significantly more expensive than Ethernet cards, and much harder to find. A quick check of a couple of recent networking product catalogs showed dozens of Ethernet cards and not a single token-ring card. The only reason to consider basing a small business network design on token ring is in deference to some prevailing concern, such as compatibility with a larger organizational network.
Asynchronous transfer mode (ATM) follows a fundamentally different approach and competes with Ethernet for backbone networks and high-performance LANs. In an ATM network, data are broken into small fixed-size cells and switched in virtual circuits established between computers. Most ATM networks operate at a very respectable 145 Mbps. Today, ATM is most commonly found as the backbone technology for enterprise networks. ATM switches are much more expensive than Ethernet hubs, and they require a significant effort to configure. Like token-ring cards, ATM cards for desktop computers are high-priced and hard to find. A small organization would use an ATM-based LAN only if it required an extremely high-performance network to support data-intensive applications such as large-scale imaging projects.
Almost all small networks will use some type of Ethernet, the most inexpensive and flexible option. Network communications catalogs are stuffed with Ethernet products from a variety of vendors. As a reflection of Ethernet's dominance, most business-class desktop computers come with Ethernet ports built directly onto the motherboard.
Ethernet is associated with a set of networking rules called CSMA/CD (Carrier Sense Multiple Access with Collision Detection), formally specified by IEEE 802.3. These network rules describe how devices on the network communicate with one another. Ethernet is a broadcast network, in which all nodes have access to all datagrams or data packets. Each packet has an origin and destination address, and each computer should open the packet only if the destination address matches its own network address. The network supports multiple devices per segment, and each device can transmit on the network at any time. If devices transmit exactly at the same time, however, a collision occurs and the transmissions are lost. Therefore, each station must check after it transmits to see if a collision occurred and, in the event of a collision, wait a random interval and retransmit.
Most varieties of Ethernet operate at 10 Mbps, and each of the nodes on a segment share this bandwidth. The stations on a segment share the overall available bandwidth and can cause collisions with one another in the process. The amount of overall bandwidth available to each station decreases and the likelihood of excessive collisions goes up as the number of stations per segment increases. The lower the number of stations per segment, the better your network will function. Various options are available to divide networks into multiple segments and to reduce the nodes per active segment.
Past their Prime: Thick and Thinwire Ethernet
There are several types of Ethernet cabling, some of which are obsolete. The original version of Ethernet, 10Base-5, or Thick Ethernet, relied on a rigid cable and required you literally to drill into the cable to install taps for each device on the network. While Thick Ethernet may still be in service in some older networks, it is obsolete and should not be used for new installations. 10Base-2, or Thinwire Ethernet, based on a thin, flexible RG-58 coaxial cable and BNC connectors, was extremely popular for a number of years because it was much easier to use than Thick Ethernet. No communications equipment was required-you just connected the network cards via the cables, and you had a functional network. Communications equipment was necessary only if you had multiple Ethernet segments that needed to be connected. The main problem with Thinwire Ethernet was its linear bus topology, where all the computers on a segment were chained together. If any single connector or cable along the segment had a problem, the entire segment would not function. Thinwire Ethernet also had limitations on the number of computers per segment and on the length of each segment; it, too, should be considered obsolete and avoided for any new network.
The primary type of Ethernet in use today is 10Base-T, which operates at 10 Mbps and follows a star topology using unshielded twisted-pair cabling.
10Base-T Ethernet networks are very easy to set up. This flavor of Ethernet relies on hubs. Each computer has a dedicated cable that connects its Ethernet card to a port on the hub. Ethernet hubs are relatively passive and require little or no configuration. In most cases, you can plug in the hub to power, connect the cable and you've got an active network. As we'll see later, there are several features to choose from when buying a hub, but almost all types are essentially plug-and-play devices.
You will need drop cables to connect your computers to the network. Pre-built Ethernet cables are available from local computer stores and mail-order companies. You can also build your own, but it's seldom worth the effort. An Ethernet drop cable must be constructed from Category 5 unshielded twisted-pair cable, and terminated with RJ-45 connectors. These connectors look much like those for a telephone jack, except RJ-45s have eight connectors instead of four.
If you are working in a small space, you may be able to connect all the computers in your network directly to the hub without putting new wiring in the walls. But in most cases you will need to have a new cabling system installed in your building to support your network.
The most common, and least expensive, devices used with 10Base-T are shared media hubs, which represent a logical Ethernet segment. Each device connected to a port on a shared media hub shares the bandwidth of a 10-Mbps Ethernet segment and competes for collisions. Multiple hubs can be cascaded together, so that all the devices on multiple physical devices still form a logical Ethernet segment. Each port on a shared media hub connects via a UTP cable to a 10Base-T interface on a network device such as a computer or printer.
When selecting an Ethernet hub, be sure to consider manageability issues. Large networks require remote management capabilities for all devices on the network; each device must be capable of communicating all aspects of its operation with a central management device through protocols such as SNMP (Simple Network Management Protocol), as well as support common implementations such as MIB-2. Most large networks will have one or more dedicated workstations monitoring the network, with the capability to monitor the status of each device, measure overall network performance and alert a network administrator when a device fails, or when its performance falls below acceptable thresholds. Ethernet hubs are classified as either managed or non-managed. If your network is large and relies on central management, it is important to purchase manageable Ethernet hubs. Managed hubs are a requirement of most enterprise networks. Typically, you can purchase one hub with management capabilities, and cascade stackable units from that hub that can rely on the base unit for management. With smaller networks, you may be able to save considerable expense by purchasing unmanaged hubs. Manageability is a relatively expensive feature. If your environment is small and does not use centralized network management, it is not cost-effective to buy managed hubs; most small office networks will work quite well without managed devices. But if you buy non-managed hubs in a managed environment, you will miss out on the ability to monitor and tune the performance of your network, and to detect and repair many network problems in time to head off a total failure.
As we noted above, a shared media hub, or a group of hubs cascaded together, represent a logical Ethernet segment. One of the advances in Ethernet technology involves the use of switching technology. Switching greatly improves both the overall performance of an Ethernet network and the bandwidth available to each station, and it minimizes the impact of errors. The major difference between a shared media hub and an Ethernet switch is that each port on an Ethernet switch is its own logical segment. A device connected to a port on an Ethernet switch has a full 10-Mbps bandwidth to itself and need not contend with other devices for collisions. No special hardware is needed on the devices that connect to an Ethernet switch. The same network interface used for shared media 10Base-T hubs will work with an Ethernet switch. From that device's perspective, connecting to a switched port is just like being the only computer on the network segment. The main disadvantage of using Ethernet switches is that they can cost several times more than a shared media hub.
One common use for an Ethernet switch is to break a large network into segments. While it is possible to attach a single computer to each port on an Ethernet switch, it is also possible to connect other devices such as a shared media Ethernet hub. If your network is large enough to require multiple Ethernet hubs, you could connect each of those hubs to a switch port so that each hub is a separate Ethernet segment. Remember that if you simply cascade them off each other directly, the combined network is a single logical Ethernet segment.
Though Ethernet traditionally has been a 10 Mbps technology, faster versions are now available. While the 10 Mbps variety continues to be the most widely implemented, 100 Mbps Ethernet is rapidly catching on. To operate at this speed, you need network cards and hubs designed for 100Base-T, both of which are now sold by many vendors. While they cost more than 10Base-T, they make a remarkable difference in performance. You can implement 100Base-T on a small network at a very reasonable cost, especially if you stick with unmanaged, non-switched hubs. Even if you choose to go with 10Base-T hubs, consider purchasing network cards that can operate at either 10 or 100 bps. 10/100Base-T cards cost only a little more than ones that operate at 10 Mbps, and give you much more flexibility for upgrading your network in the future. Most 100Base-T hubs will automatically sense whether the card connected to each port is 10 or 100 Mbps and operate accordingly. For even higher performance, you can purchase 100Base-T switched hubs. You can expect significantly higher performance with such a device, but it will also add to your costs.
This emerging fast Ethernet technology operates at 1000 Mbps. Still in its early stages, this variety of Ethernet is more suitable for connections between networks than for connecting PCs and printers on a LAN. Gigabit Ethernet is not suitable for small office networks.
When you're ready to purchase the equipment you need to build your Ethernet, you will soon see an wide variety of vendors, each one offering a range of products. Fortunately, many of the vendors have product lines that specifically focus on small business customers, such as the OfficeConnect series from 3Com, most of the Netgear products from Bay Networks and Intel's InBusiness products.
Prices for network equipment vary greatly. You can generally purchase components and significant discounts below the manufacturer's suggested retail price. Here are approximate street prices for some of the items that you'll need to build a small office network:
- 10Base-T Ethernet cards: $30-50 each
- 10/100Base-T Ethernet cards: $45-120 each
- Unmanaged Shared Media Ethernet hubs: $10-20 per port
- Switched Ethernet hub (10/100): $100-150 per port
- Fast Ethernet hub: $40-75 per port
You can purchase most of your network components from a local computer retail store, mail-order catalogs or over the Internet. Even if you purchase locally, check manufacturers' Web pages to learn which models offer the features and performance you need, and check online catalog ordering sites such as Data Communications Warehouse (http://www.warehouse.com/datacomm/) for pricing. Online ordering is often the most attractive approach. Pricing is low, the selection and availability of products is good, and most offer next-day delivery options.
Networking without a Backbone
One of the keys to designing a network for a large organization is creating methods for a large number of LANs to connect together to form a coherent enterprisewide network. The ties that bind enterprise networks include components such as hubs, routers and switches. These networks typically have a backbone that interconnects each of the individual LANs scattered throughout the organization. This backbone must be designed to move data through the network efficiently and reliably. The backbone for an enterprise network would likely use fiber-optic cabling with multiple redundant paths interconnecting the individual LANs.
The types of networks that serve small businesses follow a much simpler approach. There's no need for high-end backbone routers and switches, and all the complexity that it takes to implement and manage them. Networks for small business typically can rely on a single Ethernet hub to connect everything. In some cases, multiple hubs may be stacked or cascaded to achieve enough ports if the number of networked devices exceeds the capacity of a single hub.
After considering the various options, we can make the following recommendations for the small office network:
- Use Ethernet rather than token-ring or ATM unless there is some unusual prevailing circumstance.
- Do not use coaxial Ethernet of either the Thick or Thinwire variety.
- Support for network management generally is not required for small networks.
- Use 100Base-T if performance is the highest priority.
- Use 10Base-T if economy is the highest priority.
- Use switched Ethernet to segment the network or deliver faster performance to the desktop.
- Gigabit Ethernet should not yet be considered an option for the small office network.
Network Cabling Issues
A major part of implementing a network involves the installation of a cabling system. A solid cabling system is a good investment that will not only meet your current networking needs, but will last through your next-generation network as well.
All modern Ethernet networks follow a star topology, where each device on the network connects with its own cable to a hub or some other device. If your network is contained in a single room, then you can simply use drop cables to connect each device on your network to a hub. If your network spans an entire building, then you will need to install a cabling system, and designate one or more wiring centers for your network. Cables originate at the location of each device on the network and terminate in a wiring center. On the user end, the cable will terminate with a wallplate and in the wiring center the cable will terminate in a jack on a patch panel. To ensure that your network will work not only for your current needs, but in the future, be sure to use high-quality cables that conform to well-established standards.
In the current environment there are basically three options for connecting computers: copper wire, fiber optics and wireless technologies.
Fiber optics work very well in enterprise networks as a backbone infrastructure. Fiber offers exceptional performance for high-bandwidth applications, and is extremely reliable and secure. Fiber is not susceptible to many of the sources of interference that can play havoc with copper-based cabling systems. Fiber is also considered to be more secure since it cannot be tapped unless you cut and splice the fiber strands-a task that is virtually impossible to accomplish without detection. If you need to connect a set of buildings within a corporate complex or academic campus, then fiber optics offer the very best solution.
While it is possible to use fiber optics to connect PCs and printers in a LAN, only organizations with serious security concerns and extremely data-intensive applications regularly do so. Fiber-optic networks are expensive to implement, and their installation and maintenance demand a higher level of expertise. At a time when we can achieve 100 Mbps speed over copper cabling, it is seldom cost-effective to use fiber optics for a small office network.
Another alternative for LAN communications involves wireless technologies. By installing a wireless hub and transmitters on each computer, you can build a LAN without installing cabling at all. The cost of the wireless equipment can, in some cases, be less than the costs of deploying a cabling system, especially in buildings that lack the drop ceilings, cable trays and natural pathways needed for easy cable installation. The flexibility of moving computers around without having to deal with cable issues is another benefit. In most cases, however, cable installation costs less than putting wireless equipment in place. Wireless networks also offer less performance than cable-based alternatives. Most wireless networks on the market now operate at about 2 Mbps-roughly one fifth the bandwidth available through standard 10Base-T Ethernet, and a much smaller fraction of 100Base-T. Wireless LANs continue to be appropriate for a niche of environments where cabling issues are difficult and mobility is required. But the typical small office network will be better served by a traditional copper-based cabling system.
Most LANs are based on unshielded twisted pair (UTP) copper cabling. This cabling is relatively inexpensive and can support a great variety of network types. Most UTP cables include four pairs of copper wire, where pairs of individual wires are twisted together and the four pairs are then twisted. The quality of the wire and the number of twists determine the electrical characteristics of the cable. Properly twisted cable will be less susceptible to interference and will support high data throughput. The current standard for data cable for 10Base-T and 100Base-T networks is Category Five. Some cable installers will have Category Five-Plus or Category Six cable available. Given that the labor costs for installation are much higher than the material cost for cable, use the highest grade available. Be sure also that all jacks, connectors and patch cords involved are of the highest quality.
Should you install the cabling yourself, or hire a contractor? The smaller the office network, the more tempting it is to install cabling in-house instead of hiring professional cable installers. This will save some on installation costs, but it is important to be sure that all the cabling is installed and tested to professional standards. If your network fits within a room, then it is usually no problem to hook up the cables between the hubs. But if your network encompasses multiple rooms-especially if it is spread among multiple floors-then seriously consider hiring a professional cable installer. Make sure that the cable installer you select has specific experience with data communications networks. General electricians may not be familiar with all the requirements. An experienced data communications cable installer will know the specific grades of cables to use, will use the right connectors and will have the equipment to test the cabling system's ability to meet specifications.
Creating Network Servers and Services
In the first part of the chapter, we focused on basic connectivity issues. Now we pick up the task of creating useful network services that will help your organization use its computers more effectively. Here are some of the functions you will want to build into your network:
File services. Almost all computer applications need to store data. In a freestanding computer, the local hard drive is the primary storage device. With a network, a server can be set up to store data in such a way that it can be shared by any person in the organization. File servers are the network computers that specialize in providing shared data storage. Through the interface built into the file server's network operating system, a network administrator can set up a variety of shared folders, and control access to them. A file server will generally also offer a private folder for each user. Data stored in that folder will not be visible to any other network user. Shared folders also will be created that can be accessed by several individuals. For each shared folder, the network administrator will determine who can read its information and who is allowed to create or modify it. All the major network operating systems offer a sophisticated environment for controlling access to information stored on the file server.
Print services. Most organizations want the ability to have several individuals print to a single laser printer. It is far more efficient to purchase one or two high-end laser printers than to purchase low-end printers for each individual computer. Most network operating systems include the ability to manage network printing.
Web servers. Given the trend toward Web-based computing, you will want to consider creating a Web server for your network. Web-based systems provide efficient ways to share information, either within the company's private network or externally via the Internet.
Database and other application servers. Many businesses have specialized applications that operate from network servers. These applications may operate in conjunction with relational database systems or transaction processing environments, and are typically proprietary and industry-specific. You will need to work with the vendors that develop your applications to know the operating system, hardware and other requirements.
CD-ROM servers. Many organizations have information products on CD-ROM or DVD that need to be shared. The ability to share a CD-ROM can either be integrated into an existing file server or a dedicated server can be established.
General Network Organization: Peer-to-Peer versus Client/Server
In a network, computers tend to be considered either clients or servers. A client uses network services and a server provides them. These roles are not mutually exclusive, and some computers might operate as clients in some ways and as servers in others.
The client/server network model relies on dedicated servers that deliver services to network clients. This model makes clients and servers separate and distinct. Servers are built from larger and more powerful computer systems, and are dedicated to providing network functions. Clients rely on servers, and do not provide network services to other computers. The advantage of this approach lies in being able to concentrate resources on the server computers to ensure that they are sufficiently powerful to meet the needs of many users, have adequate security features and be highly reliable.
In a peer-to-peer network, computers can act as both clients and servers. Such a network would use non-dedicated servers that simultaneously double as client computers. Modern desktop operating systems such as Windows NT Workstation, Windows 95 and MacOS all support the ability to offer various network services in addition to providing a desktop environment. Using these operating systems, you can easily construct a peer-to-peer network where computers can function both as someone's desktop computer and offer resources shared on the network. A laser printer connected to one person's desktop computer could be defined as a network resource shared by the entire workgroup. Folders on a hard drive can be shared so others can use them to store files, and any desktop computer can also function as a Web server.
Peer-to-peer networks can be economical since they eliminate the need for high-performance dedicated servers. On a small scale, peer-to-peer networks are relatively easy to set up. But once the scale exceeds more than a handful of computers, this style of networking can be very difficult to manage and does not provide the overall computing power to meet the organization's needs. When all the shared folders reside on a centralized file server, it is relatively easy for users to know where to go to find them, and the process of determining who can and should have access is manageable. When a large number of shared resources are located on the local hard drives of computers scattered throughout the organization, the matrix of possibilities for access and storage options becomes extremely complex.
Performance issues also must be considered carefully in a peer-to-peer network. You must be sure that the computers involved have enough capacity to function well both as desktop computers and as shared network resources. The overall capacity of the computer must be adequate to serve both functions, and the competing tasks must be compatible. You could use the same computer for both your company's Web server and for your accounting department's workstation. But don't be surprised when access to the Web server slows to a crawl during tax season! Client and server functions can be combined only when the load of one of the two is relatively light.
In reality, almost all networks have some characteristics of both organizational types. Even in large enterprise networks where critical business functions operate from centrally managed dedicated servers, there likely will be some desktop computers that also function as personal Web servers, that have folders published for use within a small workgroup or that share CD-ROM drives or printers. Such peer-to-peer resource sharing within enterprise networks is often unofficial and not sanctioned by the organization's IT department.
In considering the basic setup of your small business network, carefully weigh the two options. Only the smallest networks can sustain a pure peer-to-peer approach. The more that you rely on the network for important business functions, the more you will need at least one dedicated server to guarantee fast, reliable access to network resources. If your environment at launch consists of less than half a dozen computers, then an initial network design in a peer-to-peer model might be reasonable, but plan to add a dedicated server as the network expands.
Selecting a Network Operating System
One of the most important design issues for an organization is the selection of a network operating system (NOS). This choice will affect the low-level network protocols that you will use, the way that the servers are administered and the way that users interact with network services. The NOS shapes the entire look and feel of the network environment. The NOS marketplace is very mature. The choices available, while relatively few, offer sophisticated features, high reliability, and relatively easy setup and administration.
The NOS is the software that runs on servers and defines how resources are shared on the network. It provides the basic operating environment for the server so that it can store data in a secure, reliable way, and satisfy requests from a large number of users simultaneously. Each NOS on the market today relies on extremely sophisticated multitasking, multiuser operating systems with many features designed to optimize performance, control access and ensure a high level of security. The NOS includes a set of tools the network administrator uses to configure the resources on the server as well as utilities for client computers to access those resources. At a minimum, an NOS offers file storage and network printing services. Most also offer the ability to run other specialized network-oriented applications.
Current NOS choices include Novell NetWare, Microsoft Windows NT Server and various Unix-based systems.
Novell NetWare, developed in the early 1980s as one of the first network operating systems, continues to be a major player. NetWare specializes in file and print servers. Novell networks rely on a set of underlying low-level network protocols called IPX/SPX (Internet Packet eXchange/Sequenced Packet Exchange). These protocols operate well with either Ethernet or token-ring networks, and are supported by almost all routers and other network equipment.
While widely supported, IPX/SPX today lacks the universal support enjoyed by TCP/IP. As the protocol of the Internet, TCP/IP has quickly become the dominant protocol in local networks as well. TCP/IP and IPX/SPX can operate simultaneously on a network. Many organizations use IPX/SPX for file and print service, and TCP/IP for Web-based services. Since IPX/SPX cannot be routed over the Internet, having local services based on these protocols makes them less susceptible to external attacks from hackers. NetWare is evolving toward greater support for TCP/IP. NetWare servers can run TCP/IP in addition to IPX/SPX and support Internet-style services such as Web and FTP servers. Through NetWare/IP, file and print services can operate over IP. Future versions of NetWare will embrace TCP/IP natively, allowing NetWare to operate in networks that do not allow IPX/SPX at all.
To access a NetWare server, a computer must be equipped with the proper network client software. Novell offers clients that can be installed for all major operating systems, including Windows 95, Windows NT, Windows 3.x and MacOS. Microsoft includes its own Client Services for NetWare in Windows 95 and Windows NT, so that these computers can access NetWare servers without any additional software.
In addition to file and print services, NetWare servers support other applications in the form of NetWare Loadable Modules, or NLMs. Compared to Windows NT or Unix, there are relatively few NLM applications. NetWare excels much more as a file server than as an applications server.
One of the fundamental features of the current generation of NetWare is NetWare Directory Services (NDS). This directory structure unifies all network resources into a hierarchical system. Through NDS, a network administrator defines what resources each user can access throughout the network. Novell provides a graphical application called NWAdmin that a network administrator uses to define all the resources in the NDS environment.
NDS allows users to log in once to the NDS tree and gain access to a variety of resources, even when they're physically dispersed across different servers. Previous versions of NetWare relied on a database, called the Bindery, that defined access to resources on each individual server, and users would need to log in separately to each server to gain access to its resources.
NetWare has a reputation for being extremely stable and reliable. Once set up properly, a NetWare server will operate practically indefinitely without a crash (or ABEND in Novell's terminology).
Novell NetWare was originally designed for small to medium-sized businesses, back when mission-critical functions ran on mainframe computers. Over time, the use of mainframe systems diminished and NetWare evolved to the point where it could easily support enterprisewide networks. Today, NetWare tends to be associated with large networks. Through NDS, organizations can have dozens or hundreds of NetWare servers supporting thousands of users.
Novell continues to court small-office networks. IntranetWare is a suite of products that includes the basic NetWare 4.11 operating system plus a number of applications that Novell once marketed separately. IntranetWare includes a Web server, an FTP server, NetWare/IP, the Novell Muitiprotocol Router, and the GroupWise applications for electronic mail and group calendars. IntranetWare for Small Business is a specially configured version of this applications suite, designed for easy installation and administration. IntranetWare for Small Business is much easier to install than the regular version through a set of wizards that guide the installation process. See http://www.novell.com/intranetware/products/smallbiz/ for more detailed information.
Microsoft Windows NT Server
Windows NT has steadily risen to challenge NetWare's position as the dominant NOS. It offers file and print services, and excels as an network applications server. Microsoft offers both Server and Workstation versions of Windows NT. While these products have many similarities, the NT Server has been optimized to operate as a high-performance network server and includes many security features not found in NT Workstation.
Windows NT uses a set of higher-level network protocols called Server Message Block (SMB) that operate with either NetBIOS or TCP/IP lower-level protocols. Windows NT can be easily configured to operate in a pure TCP/IP environment.
NT has become a very popular platform for network applications. Software written for DOS, 16-bit Windows or 32-bit Windows all run under Windows NT. Applications can be written to run as a native NT service to integrate fully into this environment.
Microsoft offers a package called Microsoft BackOffice for Small Business designed for the networking needs of organizations with less than 25 users. This product includes Windows NT Server, plus Microsoft Internet Information Server for Web services, Microsoft Exchange server to support electronic mail and Microsoft SQL server for database applications.
A Windows NT network uses the Domains concept to tie the network together. Multiple computers can be part of a domain, and multiple domains can exist on a network. NT version 5.0 will have a more advanced approach for organizing the network called the Active Directory. While Microsoft's domain system is generally considered less advanced than Novell's NDS architecture, it is certainly adequate for the needs of a small business network. Both Novell and Microsoft would argue fiercely about which approach to network services is superior, but either method would easily handle the challenge of organizing the small networks under consideration here.
NT servers are relatively easy to set up and administer. Any advanced user proficient with Windows 95 or with NT Workstation should be able to learn the basics of NT server with a little bit of effort. All the tools for managing an NT server are easy-to-use graphical applications.
The third main NOS alternative is comprised of the various flavors of Unix, including Solaris from Sun Microsystems. If your company uses mostly PCs, Unix does not offer the full set of services of NetWare and NT. Unix systems work well as network application servers and are widely used as Web servers and database servers. Many high-end client/server applications rely on Unix for their server.
Unix does not work well as a file server for PCs. The native means for sharing files with Unix servers involves the use of NFS or DFS, which do not come with any of the Windows or Macintosh operating systems. While NFS and DFS clients can be added to PCs through third-party applications, doing so can be expensive.
Unix has a reputation for being more difficult to install and support than either NetWare or NT. But many individuals have experience with Unix. Linux, a freeware version of Unix, has become extremely popular, and many people interested in developing their computer skills have used this operating system to gain experience with Unix.
Most small business should consider using Unix only if they have industry-specific applications that require it. Unix servers tend to be used in addition to, not in place of, an NOS such as NetWare and Windows NT. In an enterprise network, file and print services might be managed through NetWare or NT, while the organization's Web servers and database servers would be based on Unix servers. In a small network, there is a tremendous benefit to running applications servers on the same operating environment as the file servers to avoid additional system administration overhead.
Network Server Hardware
Once you have decided how to organize your network and picked an NOS, you need to select the hardware for your network server. Your organization will depend on this system for critical business functions and you want to be sure it offers adequate performance and reliability-without breaking your budget. It is possible to spend tens of thousands of dollars on a network server, but a more modest investment will likely suffice in a small network. Here are the important points to consider:
Select a server-class computer. While desktop computers are less expensive, they lack some of the features needed for full-time server operations.
Processor type. Small networks almost always use Pentium-based servers. If your organization favors Macintosh computers, then a PowerPC-based server will work well as an AppleShare server, but it does not run Windows NT or NetWare. While earlier versions of NT ran on the PowerPC, Microsoft no longer develops new versions of NT for this platform. There are also RISC processor systems to choose from such as Sun's UltraSPARC. While versions of NT that will run on this platform exist, most small business environments will stay with the Intel architecture because it is simpler and more familiar.
Processor speed. Select the fastest processor that you can afford. Faster processors come on the market with increasing frequency. The primary tradeoff is between price and performance. Buying a server based on the very latest and fastest processor will cost your company significantly more than the previous generation. If the budget for your network is tight, then purchasing the second-fastest class of processor can achieve significant savings. Otherwise, purchasing the fastest processor available will extend your investment by allowing you to make use of that server for a longer time before it becomes underpowered.
Number of processors. One of the distinguishing features of server-class computers is the ability to support multiple processors. Advanced operating systems such as NetWare, Windows NT Server and Solaris can perform computing tasks faster using multiple processors through symmetrical multiprocessing. The number of processors a server needs depends on the number of users it supports, and the intensity of the work. In large networks supporting thousands of users, it is not unusual for file servers to be equipped with four or more processors. For medium-scale applications a single high-performance processor will handle 25 users with adequate performance. As a general guideline, most small business networks can easily rely on a single-processor server. A common configuration available from many computer vendors are dual-processor systems that arrive with only a single processor installed. These allow you to operate with a single processor now and add a second processor later.
Memory. An OS thrives on a memory-rich system. Servers rely on memory not only for running applications, but also for data caching. For either NetWare or NT, install at least 64 MB of memory for a basic file server, and even more if you will be running other applications as well. Adding memory is one of the most cost-effective means to boost server performance. As your network operates over time, closely monitor how it uses memory and add more if the system becomes constrained.
Disk architecture. While IDE drives dominate most desktop computers, SCSI drives dominate server-class computers by virtue of their superior performance. The SCSI architecture makes it relatively easy to add drives or other storage devices such as tape drives and CD-ROMs.
Your file server's disk system is an extremely critical device in that it holds one of your organization's most valuable assets-its data. All reasonable measures must be taken to ensure that data cannot be lost in the event of a hardware failure, power outage, software malfunction or human error. One option for the disk storage in a file server is to use RAID (redundant array of inexpensive devices). RAID-based storage systems use several SCSI disks configured in such a way that even if one drive fails, the storage systems continue to function with no loss of data. To achieve this ability to survive hardware failures, a RAID system will have more physical disks than the amount of available storage. A significant part of the space on the disks stores parity information, which can be used to reconstruct the data on the other disks in the event of a failure. When a disk fails, data are derived from parity, and when the failed disk is replaced, its data is reconstructed. Most RAID systems will also include redundant power supplies, and software to configure and monitor the system. Most rely on a specialized disk controller. RAID systems cost significantly more than raw disk storage. The factors that would motivate an organization to use RAID would be the cost of data loss and the impact of downtime. If the nature of your environment is such that you can live with a worst-case scenario of losing one day's or part of a day's worth of data, then RAID-based storage might not be necessary. Again, small networks are the ones least likely to require an industrial-strength storage solution based on RAID.
Data backup. It is standard practice in data networks to regularly make copies of all data. Most networks rely on tape as the backup media, but devices such as optical and Jaz drives can be effective-especially for small networks. Back up your data at reasonable intervals, but do it frequently. A standard approach would be to back up all data once a week, and all new and changed files daily. This strategy ensures that no more than one day's worth of data would be lost in the event of a major failure. If the nature of your network is such that even this degree of vulnerability is too high, then increase the frequency of incremental backups to multiple times per day, or invest in a RAID storage solution.
There are numerous options for the type of media you can use for backups and how to integrate it into your network. The most common media for archiving data for backups is 8mm tape. Most 8mm tape drives are SCSI devices, and can be integrated into the file server itself or purchased as a separate peripheral.
It is just as important to have the right software for data backups as it is the right hardware. While it is possible to perform backup tasks manually, automating the process is highly desirable. The consequences of forgetting to perform a backup can be quite high. There are several automated backup systems on the market, and often a tape drive will be included with this type of software. Here are some of the features to look for in a backup system:
- The ability to automatically schedule full and incremental backups.
- Reports of all backup activities, and notification of any failed or incomplete tasks.
- Verification of archived data.
- Compression of archived data.
- Support for all file types.
- Ability to archive files that are in active use.
- Tracking of all archived files.
- Ability to selectively restore files.
- Ability to back up files on multiple servers and multiple kinds of servers (NetWare, NT, Unix).
- Ability to back up files on client workstations.
Almost all networks-even small ones-need to connect to the outside world. Internet connectivity is the most common need. If you are a branch office of a larger company, then you may need the ability to connect to the parent organization's network.
Connecting to external networks is one of the most complex parts of designing and implementing a network. Most small networks avoid the complexities of dealing with routers for their local network, but some type of routing device will be involved to connect with the outside world. Fortunately, there are some relatively simple devices available that have been specifically designed for small networks.
Most small organizations that need to connect to external networks will be able to get assistance from the group that manages the target network. In order to connect to the Internet, you will need to work with an ISP (Internet service provider) that will guide you through the process. If you are connecting to your company's home office network, there also likely will be data communications staff on board to lend a hand.
One of the most challenging aspects of implementing an external connection is determining how fast a link to install and which of the available technologies to use. The challenge lies in purchasing enough bandwidth to provide adequate performance at the lowest reasonable cost. Connectivity costs can vary from $30 per month to $20,000 a month, so you want to choose the best value for your organization's level of use.
At the low end of the market, you can rely on single computer dial-up access. With this approach, you simply install a modem in each computer and use regular analog phone lines to connect to an ISP. The main disadvantage of this approach lies in the contention for the phone lines between voice and data, and the cumulative cost of all the additional phone lines that would end up being dedicated to modem use. This model of access, though common for the one- or two-person office, is generally not sustainable for most small businesses that have at least a dozen employees. A major limitation of this approach is that it does not support the ability to provide services on the Internet. If you want to have a Web server or other resource that can be accessed on the Internet, then you need a full-time dedicated connection.
The more functional model of connecting to the Internet involves establishing a link between your local network and the Internet, rather than providing Internet access to individual computers. To enable this model of connectivity you would use a device called an router. Large networks have sophisticated routers that manage multiple internal and external connections. Such a router can easily cost $20,000 and requires an experienced network engineer to program. Small networks typically use much simpler, scaled-down devices called access routers. An access router would have an Ethernet port that connects to your local network and another port that connects to the link to your ISP and is equipped with simplified routing software that can be easily configured.
Routers are devices that help networks connect efficiently. We noted earlier that networks pass information in packets. A router examines each packet on the network and makes decisions about the quickest way to get it to its destination. Routers are protocol specific-different routing rules can be established for IP, IPX and the like. For Internet access, IP routing is all that is necessary. Even if you use IPX within your local network, you can use an IP-only access router, as long as your workstations support it. The router examines each IP packet, and if the destination address does not fall within the local network, it is forwarded to the ISP.
The world of Internet connectivity changes quickly. Until recently standard data communications links such as ISDN and frame relay have been dominant, but newer methods such as cable modems are emerging. To make a choice for your organization, you will need to explore all the options available in your area.
One very popular approach for connecting a small network to the Internet or a home office network involves the use of ISDN telephone circuits. ISDN supports both voice and data, and it relies on digital communications (standard phone service is analog). A Basic Rate ISDN line consists of two data-bearing (D) channels of 64 Kbps each and a delta (D) channel of 16 Kbps that controls the circuit. Most ISDN access routers can combine the two B channels into a single 128 Kbps datastream. ISDN also allows you to use one channel for voice calls while the other carries data. ISDN can be a relatively inexpensive way to achieve dedicated full-time Internet access, but the pricing for ISDN varies among geographical areas. Most-but not all-areas price ISDN at a flat monthly rate and do not assess per-minute access charges. If you have to pay for each minute of access, then ISDN probably will not be cost effective and you will need to consider alternatives. Again, most, but not all, ISPs support ISDN.
Frame relay is also used by many small business networks. With frame relay, you purchase a telephone circuit that connects to the telephone company's network, which in turn connects to your service provider. Options for frame relay start at about 64 Kbps, and links of 256 Kbps are quite common for small and midsize networks. Organizations with more strenuous connectivity needs will want to consider T1 or Fractional T1 links. Extremely large networks may have OC3 or OC12 links to the Internet, but these capacities are far beyond the needs of the small business network.
To connect your network to the Internet, the only relevant protocol is TCP/IP. No other protocol is routed on the Internet. Through schemes such as IP tunneling, you can connect IPX-based NetWare LANs through the Internet, but this is a relatively rare practice.
When connecting to a home office network, you have options for routing other protocols. If you are part of an organization that uses NetWare, then it is likely that you will want to implement the ability to route IPX when you establish a link to that network. Most of the midrange routers offer the ability to route IPX, but it may be an optional feature. Most low-end access routers do not route IPX. You will need to work with your network administrator to configure your router properly to manage traffic between your networks. How you assign IP network addresses and the IPX network numbers is extremely important if you plan to connect your network to external networks. In most cases the larger network will designate the various network addresses that you must use. Similar considerations hold if you're using the AppleTalk protocol with MacOS computers.
In order to have a functional network, the desktop computers in your organization will need to be connected to and configured to use the network. We noted earlier the issues involved in installing a network card in each computer, and physically connecting to the network.
The more current the operating system of your desktop computers, the less you will need to worry about network software. Windows 95 and Windows NT both come with a variety of built-in network capabilities. With Windows 3.x, you will need to install network software for both IPX and TCP/IP.
In a NetWare environment, you may want to install Novell's client rather than rely on the NetWare support built into Windows 95 or Windows NT. Novell's client offers many features not found in Microsoft's current version, including support for NDS.
The desktop computers on your network will need to be equipped with Web browsers and other network client software. As Web browsers, Microsoft Internet Explorer and Netscape Communicator both offer excellent features. The selection between these two is largely a matter of taste, though there may be some environments where one works better than the other.
As you design your network, keep a good inventory of the client computers on your network. Closely monitor the Ethernet addresses and the IP addresses you assigned to each system. This will be valuable information should you have problems and need to troubleshoot the network.
Vendor Resources for the Small Business Network
The small-office and home-office environments represent a significant portion of the network marketplace, and vendors are increasingly offering product lines and services to target these buyers. Several networking companies devote a section of their Web sites to the networking needs of small business:
Intel Small Business Home Page: (http://www.intel.com/businesscomputing/small/index.htm). Take special note of the pages titled "How Getting Connected Benefits your Growing Business" (http://www.intel.com/businesscomputing/small/running/getcon.htm).
Microsoft's small-business section can be found at http://www.microsoft.com/smallbiz. The Microsoft BackOffice Small Office Server includes a suite of applications packaged for the needs of small businesses.
Cisco, a company best known for products for large organizations and ISPs, also offers a set of products designed to connect small office networks to the Internet and other external networks. See the section entitled "Cisco Small Office/Home Office Solutions" at http://www.cisco.com/warp/public/779/smoff.html.
3Com Corp. offers a number of products for the small office market. Its Small Business Systems (http://www.3com.com/products/sbs.html) section describes the products, current promotions and general information about networking issues for these kinds of networks.
While Bay Networks specializes in solutions for enterprise networks, its Netgear subsidiary focuses almost exclusively on networking products for the small office and home environment. http://netgear.baynetworks.com/
Novell targets the small office networks with its NetWare for Small Business package. To learn more about this package and other Novell products tailored for small networks see: http://www.novell.com/intranetware/products/smallbiz/