Selecting an NIC and Network Configuration Settings

• Given a network configuration, select the appropriate NIC and network configuration settings (DHCP, DNS, WINS, protocols, NetBIOS/hostname, etc.).

Part of the role of the network administrator is to install and configure network interface cards (NICs) in client and server systems. Today, this is a fairly simple process, although you need to consider many factors. We'll start with perhaps the most basic of considerations—how to choose an NIC.

Choosing an NIC

The choice of what NIC to use depends on certain criteria, including the following:

• Bus compatibility— Some older systems have only Industry Standard Architecture (ISA) slots, but most modern systems have either Peripheral Component Interconnect (PCI) slots or both PCI and ISA slots. Either way, you should verify that there is an expansion slot of the correct type available.

• Type of network— As mentioned in the discussion on NICs in Chapter 3, “Networking Components and Devices,” unless you are using a networking system other than Ethernet, you should not need to specify another type of NIC.

• Media compatibility— As mentioned in Chapter 3, NICs can come with one, two, or even three types of network connectors.

Besides these criteria, which dictate to a certain extent which cards you can use, the choice then depends on manufacturer, cost, and requirements. The NIC might come preinstalled in the system or, as in an increasing number of cases, the network interface might be built on to the system board. In either of these situations, you do not have to install an NIC.

Installing an NIC

The physical installation of an NIC is similar to the installation of any other system expansion card. To install an NIC, you must insert it into an available expansion slot on the system board. The NIC must be pressed firmly into the expansion slot and then screwed into place. Figure 11.3 shows the physical installation of an NIC.

After the NIC has been physically installed into the computer system, it has to be configured to use the appropriate system resources. Specifically, an NIC requires a unique input/output (I/O) address and a unique interrupt request (IRQ) address. If either the I/O address or the IRQ address is the same as that of another device in the system, one or both of the devices will probably fail to work.

Windows client systems and the latest Windows server systems support plug and play, which automatically assigns system resources to expansion cards such as NICs. In non–plug-and-play Windows platforms and other operating systems, that do not support plug-and-play, the system resources are configured manually.

Regardless of the kind of NIC being used, you need driver software to make the NIC work properly with the chosen operating system. Installing driver software normally involves downloading the latest software from the manufacturer's Web site. Although the NIC might be supplied with a driver disk or the operating system might have a driver for the NIC, neither are likely to contain the latest driver, so downloading from the Web is almost certainly the best method of obtaining the driver.

Connecting the PC to the Network

With the NIC installed and functioning, the next step is to connect the PC to the network. This can be a simple process or a complicated process, depending on the type of network you are using. The following are some of the factors you should consider when connecting a new system to an existing network:

• Connecting to a coaxial network— The biggest consideration when connecting to a coaxial network is that it might be necessary to break the coaxial segment to insert a British Naval Connector (BNC) T-connector to physically connect the PC. Recall from Chapter 1, “Introduction to Networking,” that breaking a coaxial cable segment prevents any device connected to it from working. So if you are adding a computer to a coaxial segment and you need to add a length of cable and a connector, you need to either arrange with network users for a few minutes when the network will be unavailable or add the cable and connector before or after working hours. The good news is that you can leave spare BNC T-connectors in the coaxial cable segment as a just-in-case precaution. Doing so can mean that you can add a system to the coaxial segment without affecting users other than the one whose system you are connecting.

• Connecting to a twisted-pair network— Twisted-pair is the easiest of all the network types to connect to. All you need to connect is a cable (referred to as a patch cable) that connects the system to a hub or switch. In environments that use a structured cable system, the cable can be connected to a wall jack or a jack in a floor box. In a less structured environment, the cable can be run directly between the system and the hub or switch. One item worthy of note is that if you are using a Token Ring network, you must configure the NIC to work at the correct speed. Twisted-pair Ethernet networks can accommodate different speeds, if the networking hardware supports a speed higher than the base 10Mbps. Token Ring networks do not offer this function; all devices on the ring must operate at the same speed (4Mbps or 16Mbps). Connecting a system to the network with an NIC configured for the wrong speed prevents the system from communicating on the network, and it might even cause problems with other devices on the segment.

After the physical connection to the network has been established, you need to consider which other parameters need to be set. In the case of twisted-pair Ethernet networks, these parameters can include the following:

• Speed of the network— Unlike coaxial networks that operate at 10Mbps, twisted-pair networks can run at speeds of 10Mbps, 100Mbps, or even 1Gbps. Most NICs are able to automatically sense the speed of the network to which they are connected, although it is normally possible to configure the speed manually.

• Duplex settings— Recall from Chapter 3 that one of the advantages of Ethernet switches is that they allow you to use full-duplex links between the switch and the client computer. Full-duplex links allow the system and the switch to “talk” in both directions at the same time. Most modern NICs are able to automatically detect whether a full-duplex link is available and then use it. If you are using a switched network and have NICs that are able to support full-duplex links, you should make sure that this feature is being utilized.

Testing and Troubleshooting the NIC

With the NIC installed and the PC connected to the network, the next step is to test whether the NIC is functioning correctly and whether the link to the network is established.

With today's plug-and-play environment and software configurable NICs that have a range of autodetection features, testing whether the NIC is operating should be a matter of routine. You might not even have to test the NIC specifically, but just configure the NIC through the operating system and connect to the network. If all is working correctly, you should be able to connect, and by doing so prove that everything is working as it should be.

However, there is always a possibility that the installation of the NIC might have some problems. Understanding how to fix such problems is an important network administration skill.

Before considering troubleshooting the NIC at a protocol level, you should ensure that the NIC is installed and operating correctly.

There are a couple ways to do this:

• Use the tools in the operating system— Most operating systems include utilities that report the status of hardware devices installed in the system. For example, on a Windows system, the Device Manager reports the status of an installed NIC. Figure 11.4 shows an example of the General tab of the Properties dialog box for a network adapter on a Windows 2000 system.

  Figure 11.4. Information for an NIC in the Windows 2000 Device Manager utility.

  

• Use a diagnostic tool— Many NIC manufacturers provide utilities that help you test to make sure the NIC is operating correctly. The same utilities are often used to configure the NIC, as well. Figure 11.5 shows the Diagnostics tab in one such utility—the Intel PROSet utility.

  Figure 11.5. The Diagnostics tab in the Intel PROSet utility.

  

Depending on the results of the tests, you might need to further troubleshoot the installation of the NIC. If you are using a manufacturer-supplied testing utility, and it reports that it can't find the NIC, you might have the wrong utility for the NIC, or the NIC might not be working at all. Manufacturer-supplied testing utilities do not need separate driver software, so if the testing utility can't find the NIC, you can eliminate the drivers as the cause of the problem.

Configuring the NIC Settings

When you have confirmed that the NIC is operating correctly, you can configure the software settings for the NIC. The settings and configuration information you need depend on the protocol you are using.

Choosing the correct protocol is an important consideration when configuring a network or adding systems to an existing network. The client and the server must use the same protocol in order for communication to take place. This section provides a brief summary of the commonly used protocols. For a complete description of the various protocols, refer to Chapter 5, “Overview of Network Protocols.”

• TCP/IP— By far the most prevalent of network protocol suites, TCP/IP is available for almost every computing platform and has widespread industry support. The majority of LANs now use TCP/IP as the default protocol. Configuring TCP/IP connectivity requires the use of an IP address, a subnet mask, a default gateway, and possibly Domain Name Service (DNS) server information and Windows Internet Naming System (WINS) information.

• IPX/SPX— Novell invented and implemented IPX/SPX when it introduced NetWare in the 1980s. At that time, TCP/IP was for the most part an academic/military/government protocol, and Novell realized the need for a robust, routable protocol. IPX/SPX is one of the main reasons that Novell owned the networking market through the 1980s and most of the 1990s. IPX/SPX was also easy to install and configure. Today, TCP/IP has largely displaced IPX. One of the advantages of IPX is that workstation configuration is very simple. Generally speaking, the only item that might need to be configured is the frame type, which determines the format in which data is grouped into the frames that are placed on the network. Older versions of NetWare use a frame type called 802.3, whereas newer versions use a frame type called 802.2. Fortunately, most client software is able to detect the frame type automatically.

  IN THE FIELD: A NEW LEASE ON LIFE FOR IPX? With the advent of the Internet, IPX/SPX has found a new home as a security tool. To ensure that internal network resources are not exposed to unauthorized people from external networks, such as the Internet, some administrators are using IPX/SPX on the internal network. IPX/SPX provides a routable protocol, allowing the construction of large, efficient internal networks that are automatically isolated from the outside world. Administrators can use hardware and software implementations that translate information to and from the Internet. Many administrators like this extra level of security for their internal networks. You can consider using IPX/SPX in NWLink and NetBEUI network environments.

  
  

• NWLink— When Microsoft began working on adding support for interoperability with NetWare, it opted to develop its own fully compatible version of Novell's proprietary IPX/SPX. This development was necessary because earlier versions of NetWare did not support authentication over TCP/IP.

• NetBEUI— Microsoft chose IBM's NetBEUI as the protocol for its first networking implementation in the mid-1980s. One of the reasons Microsoft chose to base its early networking efforts on NetBEUI was the protocol's simplicity and speed. Microsoft wanted to offer a very simple, easy workgroup configuration. Name services and addressing are both handled automatically with NetBEUI. There are no configuration issues, other than setting up the NIC and installing NetBEUI as the protocol. Because of NetBEUI's simplicity, administrators sometimes use it to troubleshoot hard-to-find communication problems between two machines. The simplicity of NetBEUI also created problems for Microsoft as the 1980s progressed. NetBEUI is a nonroutable protocol, and as networks began to interconnect, Microsoft found its clients stranded within the confines of small LANs.

As mentioned earlier, TCP/IP is by far the most common of the networking protocols in use today. For that reason, we will take a more in-depth look at configuring client systems to use TCP/IP.

Configuring Client Systems for TCP/IP

Configuring a client system for TCP/IP can be a relatively complex task, or it can be simple. Any complexity involved is related to the possible need to configure TCP/IP manually. The simplicity is related to the fact that TCP/IP configuration can occur automatically via DHCP. We'll discuss DHCP later in this chapter; this section looks at some of the basic information required to make a system function on a network, using TCP/IP. At the very least, a system needs an IP address and a subnet mask. The default gateway, DNS server, and WINS server are all optional, but network functionality is limited without them. The following list briefly explains the IP-related settings used to connect to a TCP/IP network:

• IP address— Each system must be assigned a unique IP address so it can communicate on the network.

• Subnet mask— The subnet mask allows the system to determine what portion of the IP address represents the network address and what portion represents the node address.

• Default gateway— The default gateway allows the system to communicate with systems on a remote network, without the need for explicit routes to be defined.

• DNS server addresses— DNS servers allow dynamic hostname resolution to be performed. It is common practice to have two DNS server addresses defined so that if one server becomes unavailable, the other can be used.

• WINS server addresses— A WINS server enables Network Basic Input/Output System (NetBIOS) names to be resolved to IP addresses. As with DNS servers, it is common practice to enter two WINS server addresses, to provide a degree of fault tolerance.

Exactly how this information is entered on the client depends on the operating system being configured. For example, Figure 11.6 shows the Internet Protocol (TCP/IP) Properties dialog box on a Windows 2000 system. As you can see, the system represented in Figure 11.6 is fully configured for operation on a private network.

The configuration screens for other systems are slightly different. Figure 11.7 shows the IP Address tab of the TCP/IP Properties dialog box on a Windows Me system.

In any case, the parameters required need to be entered into the respective dialog boxes carefully. In the case of Windows Me, the DNS Configuration and WINS Configuration tabs must be used to input the DNS and WINS information. In Windows 2000, the DNS server fields are on the same screen as the main IP address.

Other systems use different utilities to allow TCP/IP configuration information to be entered. Figure 11.8 shows the TCP/IP configuration screen on a Macintosh OS9 system, and Figure 11.9 shows the TCP/IP configuration screen in the netconfig utility on a Red Hat Linux system.

Configuring DNS Server Information

DNS is used on TCP/IP networks for name resolution. It resolves fully qualified domain names (FQDNs) to IP addresses. For example, DNS would resolve the address www.comptia.org to its IP address, 216.119.103.72.

Regardless of the operating systems used, at least one DNS server must be accessible by a client system in order for dynamic name resolution to take place. Clients on the TCP/IP network must be configured with the IP address of the DNS server. Figure 11.10 shows the DNS configuration tab of the Advanced TCP/IP Settings dialog box on a Windows 2000 system. (You can find more information on DNS in Chapter 6.)

Configuring WINS Server Information

WINS is used to convert NetBIOS names to IP addresses. NetBIOS names are the friendly names by which we refer to the computers on the network (for example, sales1, Maryscomp, secretary).

Many client operating systems are WINS enabled, which means they can be configured to use WINS servers. WINS-enabled clients use the WINS server to resolve NetBIOS names to IP addresses, allowing communication across subnets and reducing broadcast traffic. The following is a partial list of the client systems that can use WINS:

• Windows 2000 Server/Professional

• Windows NT Workstation/Server

• Windows 95/98/Me

• Windows for Workgroups

• OS/2

• Unix/Linux (requires Samba)

Each of the WINS-enabled client systems needs to be configured to use WINS. To do so, the IP address of the WINS server is required. Figure 11.11 shows the WINS Configuration tab of the TCP/IP Properties dialog box on a Windows Me system, and Figure 11.12 shows the WINS Configuration tab of the Advanced TCP/IP Settings dialog box on a Windows 2000 system. (You can find more information on WINS in Chapter 6.)

Using DHCP

Now that we have discussed how TCP/IP configuration information is entered manually, it's time to kick back a little and look at how you can handle this configuration with a single click—thanks to DHCP. As mentioned previously, DHCP is a protocol that is used to simplify the assignment of IP configuration information on a TCP/IP network. DHCP allows you to dynamically assign IP addressing information to client systems on the network, reducing possible human error and administrative overhead. DHCP is not restricted to a single platform; it is a generic technology that is supported by all the major operating systems.

When DHCP is used on a network, the client systems must be configured to use DHCP. Each of the client operating systems has some method of configuring the system to receive IP information from the DHCP server. Configuring a client to use DHCP is often as simple as clicking a check box or selecting a radio button.

When client systems are configured to use DHCP, they can receive more than just the IP address. They can also be assigned any of the other TCP/IP information, such as the default gateway, subnet mask, and any DNS or WINS servers. In addition, some DHCP server platforms support a range of other information as well. (You can find more information on DHCP in Chapter 6.)

When the networking configuration is complete, the system should be able to function on the network. However, to connect to a server system and use its resources, the system needs client software, which is discussed in the following section.