'; window.popUpWin.document.write(zhtm); window.popUpWin.document.close(); // Johnny Jackson 4/28/98 } //--> Windows 98 Professional Reference -- Ch 25 -- Windows 98 with TCP/IP


Windows 98 Professional Reference

Previous chapterNext chapterContents

- 25 -
Windows 98 with TCP/IP


The rise of the Internet and the recent emphasis on interoperability have only increased the stature of TCP/IP. The TCP/IP protocol suite, which grew up around UNIX, is quickly becoming the universal network protocol, and Microsoft provides some new and innovative features with Windows 98's TCP/IP. If you're accessing the Internet, you'll need to use TCP/IP. If your Windows 98 computer will be part of a routed, non-Novell LAN or WAN, the chances are you'll need TCP/IP. Windows 98's new Automatic Private IP Addressing feature (APIPA) makes it easier than ever for even small workgroups and single-subnet local networks to use TCP/IP.

This chapter discusses some basic TCP/IP configuration issues and shows how you can configure Windows 98 to operate on a TCP/IP network.

TCP/IP Concepts

TCP/IP, like any other networking protocol, is a system of rules that facilitates communication among computers. An implementation of TCP/IP, such as Windows 98's TCP/IP implementation, is a software component that carries out the tasks associated with communication through the TCP/IP protocol. Chapter 21, "Understanding Windows 98 Networking," discussed the basics of protocols and protocol bindings in Windows 98 and described how to configure network protocols using the Control Panel Network application. TCP/IP generally requires more configuration than other protocols, and this chapter will help you understand some of those configuration options.

TCP/IP is often referred to as a "protocol suite" or a "protocol stack" because it is a collection of related protocols and applications that provide Wide Area Networking (WAN) and Local Area Networking (LAN) functionality.

The TCP/IP suite includes protocols used to communicate at the network level, protocols used to manage the transfer at the internet level, and the application programming interfaces provided at the application level of the network model shown in Figure 25.1 below.

Figure 25.1

The TCP/IP network model - broken down by layers.

Every computer in a TCP/IP network has a unique IP address. An IP address is a logical address identifying the computer on the network. The IP address is a 32-bit binary number. To make the IP address more memorable and less intimidating, the 32 bits are subdivided into 4 sets of 8 bits, known as octets. The IP address almost always appears in decimal form. In decimal form, each octet is represented with a number less than or equal to 255 (which is 28-1). The four octets are separated by periods, producing an address such as:

12.234.5.67 (read as twelve dot two thirty-four dot five dot sixty-seven)

NOTE: The ARP protocol translates IP addresses to Ethernet or Token Ring physical addresses for the subnet. The physical address is a permanent address associated with each network adapter. For more on ARP, see the discussion of the ARP utility, later in this chapter.

The IP address consists of two parts: a network ID and a host ID. All computers on any given network segment will have the same network ID. You'll learn more about IP addressing later in this chapter.

Windows 98 supports three methods for aquiring an IP address. Choosing a method to use will depend on your network and your needs. The IP address configuration methods as follows:

These three address configuration methods are each discussed in later sections. You must assign a legal and appropriate IP address to your computer if you want it to operate on a TCP/IP network.

Referring to computers in the form of binary 32-bit octets is convenient for computers, but it is less convenient for flesh-based devices such as humans. Computers on a TCP/IP network typically are referenced by a human-friendly name that is then resolved to an IP address by the computers on the network. Windows 98 supports two schemes for assigning names to computers, as follows:

These name-resolution schemes are discussed later in the chapter.

Another important concept you'll need in order to understand TCP/IP networking is a gateway. The term gateway has many definitions within the world of networking, and recently has come to mean a device that performs some form of protocol translation. Within TCP/IP, however, a gateway is simply a router that acts as a conduit from the subnet to the greater network. The default gateway (which you can specify through Windows 98's TCP/IP settings) is the default location where Windows 98 will send packets addressed to destinations beyond the subnet.

TCP/IP is designed to serve very large networks. Indeed, the Internet (the largest TCP/IP network) covers the entire planet and is many times larger than anyone ever imagined a network could be even a few years ago. To help manage and troubleshoot traffic over the great spaces of the network, TCP/IP comes with a number of useful utilities. Some of these utilities are also useful on small networks. A later section of this chapter discusses their TCP/IP utilities.

Configuring TCP/IP

As mentioned earlier, there are three methods for assigning a TCP/IP address to a Windows 98 machine, as follows:

As is so often the case, the manual method is the easiest to understand conceptually, but the most time-consuming to implement (especially on a network) because it requires you to directly configure an IP address and subnet mask for each network node. The manual method (and the dynamic method, if you're using an onsite DHCP server) require you to have some conception of what you'd like the IP address to be. The IP address for a given computer must fit into a coordinated addressing scheme for the complete network (see "Understanding IP Addressing," later in this chapter). The automatic method, which uses Windows 98's new APIPA feature, does not require any knowledge of IP addressing techniques and is thus ideal for small networks with little or no on-site technical administration.

Configure TCP/IP properties through the Network Control Panel. You must first make sure TCP/IP is installed and bound to the network adapter through which you wish to connect to the network. (See Chapter 21 for more on installing and binding network protocols.) To access the TCP/IP Properties dialog box (see Figure 25.2), open the Network Control Panel and double-click on the icon showing TCP/IP bound to the relevant network adapter.

Figure 25.2

The TCP/IP Properties dialog box.

The TCP/IP Properties dialog box includes seven tabs, as follows:

If you're connecting to the network through a dial-up adapter, as would be the case if you're configuring dial-up access to an Internet provider, Microsoft recommends that you configure TCP/IP settings through the Dial-Up Networking connection rather than through the Network Control Panel. (See Chapter 26, "Windows 98 and Remote Communication," for more on Dial-Up Networking.)

Dynamic IP Address Assignment with DHCP

The Dynamic Host Configuration Protocol (DHCP) is a centralized method for providing IP addresses and all or most of the associated other IP setup information for a client. When the client is first booted, a Discovery packet is sent on the local network. DHCP servers listen for these Discovery packets and send an Offer packet back to the requesting host. The host then responds with a yes to the first offer, no to all other current offers. The DHCP server then sends a final Acknowledge packet to the host. After this point in the process, the client will have a valid IP address and all associated options, like WINS server(s), DNS server(s), Node Type, Scope ID and or Router IP, will also be configured.

If a certain machine needs to have a permanent IP address, then the DHCP service can be configured to always provide the same IP address for that machine based on its Media Access Control (MAC) address.

To set up Windows 98 for DHCP, select the radio button labeled Obtain an IP Address Automatically, in the IP Address tab of the TCP/IP Properties dialog box (as described earlier in this chapter).

Automatic IP Address Assignment with APIPA

Windows 98's Automatic Private IP Addressing (APIPA) feature is designed to ensure that your computer will still be able to connect to the network if an IP address isn't available. APIPA has two primary functions, as follows:

At startup, if your Windows 98 computer isn't manually configured with an IP address, it looks for a DHCP server. If it can't find a DHCP Server, it does one of the following:

The Windows 98 computer then communicates using the Net 10 (10.X.X.X) address format. APIPA machine will not be able to use WINS or DNS, and they will not be able to communicate with nodes that are inaccessible through Net 10 addressing.

If A DHCP server comes back online, the APIPA computer accepts a new leased IP address and discontinues the APIPA address.

APIPA is a viable option for small, simple networks (fewer than 10 nodes, according to Microsoft), but for larger networks or for subnetted networks, APIPA is not a permanent solution.

Name Resolution

Windows 98 can employ several different methods for resolving host names and computer names into IP addresses (and resolving IP addresses into names). If Windows 98 encounters a name where it might have expected an IP address, it checks the following:

Local Host Name. Is the name requested the same as the current host name?

HOSTS file. Look through the HOSTS file, which provides IP-to-DNS-name mappings

DNS. Send a request to a DNS server if one is configured

WINS. Send a request to a WINS server if one is configured

Broadcast. Send a broadcast on the local network to resolve IP-to-computer-name queries

LMHOSTS file. Look through the LMHOSTS file, which provides a static table of IP-to-computer-name mappings

A summary of these name-resolution techniques is as follows:

Native Windows networking features, such as Network Neighborhood, use Windows computer names to support network browsing. If your network is using TCP/IP, you must provide some form of IP-to-computer-name resolution if you wish to support browsing across routers.

NOTE: A Windows NT Server system can act as a DNS Server or a WINS server.

The following sections describe how to configure Windows 98 to use a DNS server, a HOSTS file, a WINS server, or an LMOSTS file.

Setting Up Windows 98 for a DNS Server

To set up a Windows 98 machine to use DNS, select the DNS Configuration tab in the TCP/IP Properties dialog box and make the following entries:

1. Select the radio button to Enable DNS.
2. Although the Host field may be any legal name, the default host name would normally be the computer name. You can put in another name here to use as the host name without affecting the computer name. The host name is used together with the domain name to create what is known as a Fully Qualified Domain Name (FQDN) for the machine. When the machine does a DNS query, the local domain name is appended to the short name (host name) to create a FQDN.

NOTE: The DNS domain name is unrelated to the Windows NT domain name. You'll configure a DNS domain name when you set up a DNS server. If you want your network to be available by host name over the Internet, you'll need to register a domain name with InterNIC: http://rs.internic.net.

3. Put the domain name in the Domain name box. This is the name of the Internet domain and is not the same as a Windows NT or LAN Manager domain. NT and LAN Manager domains are logical groupings of computers for administrative purposes. This would be a domain name such as mcp.com. As stated in step 2, this information will be appended to the host name to generate FQDNs for query resolutions. The domain name field is optional, but the more information you provide, the fewer problems you are likely to encounter.
4. In the DNS Server Search Order box, type in the IP address of up to three DNS servers that you want to use for name resolution. The order in which these servers will be searched will depend on the order in which you type them. The first one in the list will be the first one searched and so on (see Figure 25.3 for an example). Once you have typed in the IP address of the DNS server, click the Add button to add it to the list. If you need to rearrange the order in which the severs are searched, you will need to manually reconfigure them by removing and adding them until the list is in the order you desire.
5. If you would like to have additional domain suffixes appended to the host name and searched in queries, type in the names of those additional domain suffixes to the Domain Suffix Search Order box and click Add to add them to the list. You may enter up to five additional suffixes to search.

If you plan on using WINS in Windows 98, then you will also need to configure the WINS configuration tab. WINS reduces the amount of local broadcast traffic done for name resolution. WINS can be used alone or in conjunction with DNS. WINS also enables the users on the network to browse across the routers without having to have a special configuration file for each machine.

Figure 25.3

Configuring the DNS Configuration tab for the TCP/IP protocol.

Setting Up Windows 98 for a HOSTS File

HOSTS Files

In the early days of TCP/IP, the HOSTS file was used to translate known DNS host names into their equivalent IP address. The HOSTS file is a simple text file with one line for each record. A record consisted of one IP address followed by one or more host names to be used for that IP address. For example, a sample of a host file may look like the following:

Table 25.1 Sample HOST File

199.72.26.10 callen CALLEN callen.mis # Win 98
199.72.26.11 rakesh RAKESH rakesh.fin # P200 NT
199.72.26.12 cindy CINDY cindy.fin # Alpha VMS
199.72.26.13 sarah SARAH sarah.accts # P200 Linux

In Table 25.1, above, the host names cindy, CINDY, and cindy.fin all refer to the IP address 199.72.26.12. The other names can be used to indicate groupings, computer types, locations, or anything else you desire.

See the sample HOSTS file HOSTS.SAM in the \Windows directory for a discussion of HOSTS file format. You can edit the HOSTS file using any text editor, such as Notepad.

Because the HOSTS file must be manually configured, you'll need to continually update the file whenever there is a change to an address. The HOSTS file goes in the \Windows directory; it has no extention.

NOTE: You must select the Enable DNS option in the DNS Configuration tab of the TCP/IP Properties dialog box in order to use a HOSTS or LMHOSTS file.

Setting Up Windows 98 for a WINS Server

To set up WINS, select the WINS configuration tab and make the following entries:

1. Click the radio button to Enable WINS Resolution.
2. In the WINS Server Search Order box, type the IP addresses of the Primary and Secondary WINS servers. Click the Add button to add them to the list.
3. You can enter a scope identifier in the Scope ID box if you will be using NetBIOS over TCP/IP and require it on your network for a group of computers to communicate with each other and not outside their group.
4. If you will be using DHCP and have configured the IP Address tab to automatically assign the IP address, then the radio button for Use DHCP for WINS Resolution will be available to select. If you select this option, then you may leave the rest of the fields blank and have the DHCP server provide your Windows 98 machine with the information it needs.

NOTE: Don't forget that if you have more than one adapter in your Windows 98 machine, you may have to configure these items more than once or possibly with different settings for different adapters.

Setting Up Windows 98 for an LMHOSTS File

The LMHOSTS provides static mapping of computer names to IP addresses. Like the HOSTS file, LMHOSTS must be updated manually. Using LMHOSTS is therefore less convenient and not as versatile as using a WINS server for name resolution, but in some situations, an LMHOSTS file is still a viable option.

See the sample LMHOSTS file LMHOSTS.SAM in the \Windows directory for a discussion of LMHOSTS file format. You can edit the LMHOSTS file using any text editor, such as Notepad. The LMHOSTS file goes in the \Windows directory; it has no extension.

TCP/IP Tools

The TCP/IP suite is full of useful user and administrative tools. Most of the tools are run from the command line. This means you must run them from the command prompt or, in some cases where there is a GUI display, the RUN command in the Start menu, or double-click the executable in Explorer or My Computer.

WINIPCFG

The Windows 95/98 IP Configuration Display utility provides a basic view of the IP configuration. If the More Info >> button is clicked, then a complete listing of options appears as in Figure 25.4 below. This figure shows a basic configuration of a host with IP and Mask. This display also displays the MAC address of the network interface card. This host has been set as Broadcast Node Type since no other methods of host name resolution are configured, like DNS or WINS.

Figure 25.4

WINIPCFG Utility - Basic Host Configuration of IP and Mask value.

Telnet

Windows 98 provides a TELNET client application. This is a basic terminal emulation package providing connectivity to any terminal based operating system like UNIX and VMS. A terminal emulation provides interpretation of standard escape sequences that indicate cursor movement, cursor position, screen color, screen reverse, screen flashing, and terminal beep operations to name a few.

The client can be used to connect to any host. If you are connected to the Internet or have an internal DNS service, you can use a fully qualified domain name, FQDN, to access the host. If you have WINS, you can use the NetBIOS name. If you have neither WINS nor DNS, you may need an LMHOSTS file or a HOSTS file with host name information in it for every host you want to access by name. If all else fails, you can use the IP address of the host, no matter what, assuming, of course, you know it.

To use TELNET (see Figure 25.5), at the command prompt, enter the command with the host name or IP address as an argument such as:

    TELNET    HOSTNAME.DOMAIN.ORG
    TELNET    NetBIOSName
    TELNET    XX.YY.AA.BB

You can also create a shortcut to TELNET on your desktop or in your Start menu settings, per Windows 98 standard methods. With a shortcut, you can include the IP address of a specific host if desired, otherwise it will ask for the host or IP address to connect to. Remember the host names must have some resolution method available, otherwise only the IP address will work.

Figure 25.5

The TELNET Application; Connecting to the world, connected to a TELNET Daemon running on NT, at an NT CMD.EXE prompt.

FTP

Another standard utility supplied with TCP/IP is the client application FTP, or File Transfer Protocol. This is similar to a TELNET session except that it has some specific commands for retrieving, (GET, a file) and uploading, (PUT, a file). You can also do multiple file transfers with MGET and MPUT commands within FTP.

Like a TELNET session, an FTP session is terminal based and requires validation at the server host. You need to know an account name and password at the FTP server to gain access.

The FTP protocol was one of the building blocks of the Internet. To allow anyone access to an FTP site on the Internet, a special login name was used called anonymous. You need to supply your email address as the password for anonymous access.

To use the FTP utility at the command prompt, like TELNET (see Figure 25.6), enter the command with the host, FQDN, or IP address of the FTP site to visit. For example:

FTP    FTP.CLASSIC.COM

FTP    206.195.1.2

This would connect you to the CLASSIC.COM FTP server host, assuming the name can be translated to a valid IP address by some service, HOST file, LMHOST file, or DNS service.

Figure 25.6

The FTP Client--Sample connection (to NT IIS).

PING

The first tool of TCP/IP is the connectivity tool known as PING (see Figure 25.7). PING sends packets from one host asking for a return of the same information. A successful PING of another host provides delivery timing information and proves all software and hardware on both machines is functioning properly.

Figure 25.7

PING Utility from the DOS Command Prompt-- defaults to 4 test packets.

ROUTE

The ROUTE command can be used to display the current routing table, as well as add, modify or delete any routes in the table. You only need to add a route if you have multiple network interfaces.

Windows 98 has been upgraded to provide basic routing functionality. This requires two network interfaces on a single host. The concept of routing is that the software at one interface receives a packet destined for the other interface. The software then transfers the packet to the other interface with all required IP packet header changes for the new network. This is called IP Forwarding.

To enable the actual routing service, you would need to click the Enable Routing check box in the WINIPCFG utility as shown in Figure 25.8.

Figure 25.8

ROUTE Command Sample Output after last part of help output.

TRACERT

To test your wide area network connectivity, you can use TRACERT at the command prompt. This command actually sends out PING commands with incrementing by 1 Time-To-Live (TTL) values starting with 1. This forces the first router to send an Internet Communication Message Protocol, ICMP, error message back. TRACERT displays the IP and possibly the name of each target along the path to the remote host by incrementally forcing the next host after each error response until the actual destination host responds.

You will only need this on the rare occasion that the network response is slow or dead. You can see at which point along the path the congestion is occurring. The average user will never need to use this utility.

Other Utilities

The NetBIOS Over TCP/IP statistics are available using NBTSTAT at the DOS command prompt. The information is time specific in that the entries only last between 2 and 10 minutes after their last use. When a NetBIOS request is made and the destination machine name is unknown, your system broadcasts a request for the name on the local network. If a machine answers, with its MAC address, then this MAC address is stored in the NetBIOS cache for use by the protocol to transfer data between the source and destination. NBTSTAT displays the names and their MAC addresses in this table.

The only reason you might use this utility is to troubleshoot a connection that is not correct or not working. You would check the NBTSTAT entries with NBTSTAT -N and compare the results with the MAC hardware address of the correct machine. The average user will not need to use this utility.

Understanding IP Addressing

Every computer on a network must be uniquely identified so that network applications will know where to send data. On a TCP/IP network, this unique identification comes in the form of an IP address. In TCP/IP, every connection to the network must have a valid IP address. While it's tempting to think that a network device (such as a computer, router, or printer) has the network address, this is not the case. It is actually the device's connection to the network that has the address. Some devices (routers, for example) will have more than one connection to a network and will need more than one IP address. In this section, we'll look briefly at how IP addressing works.

The IP Address Format

If you've used a TCP/IP-based network (and the Internet qualifies), you've probably seen IP addresses. They typically take a dotted decimal form, such as 192.102.11.103--four numbers ranging from 0 to 255, separated by decimal points. What you may not know, however, is that this dotted decimal notation is really just a human-friendly way of representing the address's true form, which is as a 32-bit binary number. This 32-bit number is broken down into four 8-bit segments called octets. Each of these octets ranges from 00000000 (0 in decimal) to 11111111 (255 in decimal). Written out in binary format, the IP address example I gave earlier (192.102.11.103) looks like this (I added the spaces to make it easier to look at):

11000000 1100110 00001011 1100111.

Network and Host IDs

Every IP address contains two parts: a Network ID and a Host ID. The Network ID defines the network segment on which a particular connection exists. The Host ID identifies that connection on that network. Think of it this way. Your house has a street address, something like 123 Main Street. That street address is like an IP address and can also be broken down into two parts. The street name, Main Street, is like the Network ID in an IP address. The street number, 123, is like the Host ID in an IP address and identifies the house on that street.

In an IP address, however, things are the opposite way around from a street address. The Network ID is on the left and the Host ID is on the right. Take the IP address we used earlier, 192.102.11.103. The left two octects (192.102) might be the Network ID and the right two octets (11.103) might be the Host ID. You probably noticed I used the word might. That's because where the split occurs in the IP address can change. This leaves room for flexibility in assigning IP addresses.

Address Classes

The Network Information Center (NIC) is responsible for assigning IP addresses. They have three classes of address that they can assign: A, B, and C. Class A networks are assigned to big companies, Class B to medium sized-companies, and Class C to smaller companies. Class A and B addresses are no longer available. They have all been given out. The way that these classes are determined is where that split between network ID and Host ID occurs. Take a look at table 25.2.

NOTE: that the terms Network Information Center and Network Interface Card have the same abbreviation (NIC).

Table 25.2 NIC Network Classes

Class Format First Octet Networks Hosts
A N.H.H.H 1-127 127 16 million
B N.N.H.H 128-191 16,384 65,534
C N.N.N.H 192-223 2 million 254

The first column in Table 25.2 shows the class of the network. The second shows how many octets of the IP address are assigned to be the Network ID and the Host ID. The class defines a default subnet mask, and the subnet mask defines how the IP address is divided between the Network ID and the Host ID. You can also customize the subnet mask, as described later in this chapter. In a Class A network the first octect is the Network ID and the last three octets are the host ID. This means that there are a relatively small number of class A networks (127), but that each can have a large number of hosts (over 16 million). Finally, the table shows what the decimal range of the first octect is for that class. Class A networks range from 1-127. In a class B network, you can see that the first two octects are used as the Network ID and the last two as the Host ID. This means more networks (16 thousand), but fewer hosts per network (only 65,000).

NOTE: Addresses beginning with 127 are actually reserved for internal use and cannot be used for IP network addresses. This address is referred to as the local host and is commonly numbered 127.0.0.1 although any valid octal numbers after the 127 are considered the same in most cases.
The first octect range from 224-239 is used for multicast addresses, allowing hosts to broadcast to multiple destinations. The range from 240-255 is used for experimental purposes.

Subnets

Often, your network will be divided into smaller networks called subnets. This division can be useful for many reasons, including the reduction of network traffic and the simplification of network management. But, what do you do if you need to subnet your network and you've only been given one network address? TCP/IP allows for a nice solution to this problem, called subnetting. Subnetting allows an IP address to be broken down into smaller, logical subnets. Basically, what happens is you take the original Host ID of the IP address and break it down into a subnet address and host address. Referring back to our example of the street address, this would be akin to breaking down a street number into apartments or suites.

Subnetting works something like this. Assume that your company has been assigned a Class B address by the NIC. This address is 131.107.x.x. Remember that a class B address typically uses the first two octets as the Network ID. In this example, therefore, the Network ID would be 131.107. As is, we have one network which can have just over 65,000 hosts on it, each host identified on the network by the second two octets of the IP address. That's fine if we want a network that large. We have another option, though. We can take the Host ID (the third and fourth octect) and make one the subnet ID and one the Host ID. Now we have one big network (represented by 131.107.x.x), which is divided into 254 subnetworks (131.107.1-254.x), each of which can have 254 hosts (131.107.x.1-254).

Subnet Masking

The idea of subnetting is pretty simple, but how do we actually do it? Subnetting is accomplished using something called a subnet mask. A subnet mask is another 32-bit binary number that identifies which part of an IP address is the network ID and which part is the Host ID. In binary, the subnet mask will always be a string of ones followed by a string of zeros, like this:

11111111 11111111 00000000 00000000

The shift between one and zero indicates where the network ID of the IP address the subnet mask defines ends and where the Host ID begins. Like the IP address, the subnet mask is often represented in dotted decimal notation. Thus, the example above would be 255.255.0.0.

Let's take a look at an example. Assume that I have a standard Class B address, like 131.107.34.123. The default subnet mask for a Class B address is 255.255.0.0. When I give a host the address 131.107.34.123 and the subnet mask 255.255.0.0, the first two octets (131.107) are used as the network ID and the last two (34.123) are used as the host ID. If I decide to subnet my network, as in the example we discussed earlier, I might use the subnet mask 255.255.255.0. Now, the first three octets (131.107.34) are used as the network ID and only the last (123) is the host ID.

So far, we've looked at subnet masks in which each octect is either 255 (11111111 in binary) or 0 (00000000 in binary). You can actually make the division at any point after the default division used for your network class. This allows you to customize the number of subnets and hosts per subnet to more precisely meet your situation. Using the previous example again, you might decide that instead of having 254 subnets and 254 hosts per subnet, you may need more hosts on each subnet. You could create a custom subnet mask which would allow just that. It's easiest to understand this if we look at it in binary. To get 254 subnets and 254 hosts, we used the following subnet mask:

11111111 11111111 11111111 00000000 (255.255.255.0)

To create a situation where we have more hosts per subnet, we would simply increase the size of the Host ID by moving the subnet mask to the left. We might, for example, use a subnet mask like this one:

11111111 11111111 11110000 00000000 (255.255.240.0)

Developing Your Network

To develop addresses for your own TCP/IP network, first, figure out how many clients you will need to accommodate at an absolute maximum for any one area and then start with the network class that supports at least that many hosts. For example, if there is less than 254 hosts in any network, use all class C network numbers, just make them up like 222.1.1.0 is the first network all the way up to 222.1.254.0 is the 254th network. That is a total of almost 64,000 possible IP addresses, 254 networks each with 254 hosts, to choose from. This should be enough in most cases. Use the default mask of 255.255.255.0 for each network.

If you need more than 254 in any one site and no more than 65,000 then use a class B network address of your choice. For example, you could use 134.1.0.0 for the first network and 134.254.0.0 for the last network, providing 254 networks of 64,000 clients per network, whether you use them all or need them. Use the default mask of 255.255.0.0 for each network. In this case your client numbers for any one given network range from x.y.0.254 to x.y.255.254.

Hopefully, this has given you a little understanding of how IP addressing works. Obviously, there's more to it than I can cover in a section like this one.

SNMP and Windows 98

The Simple Network Management Protocol, or SNMP, is another specialized protocol used to monitor and manage hosts remotely. Windows 98 comes with an agent. An agent is a service that accepts specific requests for information about various parameters on this host. The agent allows a controlling SNMP management utility, like Tivoli, CA Unicenter, Openview, Netview and many more, to `watch' over each SNMP client on the network. The manager can set counters to specific values or reset them to zero, set threshold values that will cause the local agent to send an alert or trap back to the management software in the case of an error. The SNMP agent provides a simple method of watching and tracking all nodes on the network that have SNMP agents. Almost every network device built within the last few years will be SNMP agent equipped.

To install a local SNMP agent for Windows 98, you need to install it from the Add Services in the Services panel of Network Icon from Control Panel. The software is located on your installation CD-ROM in:

E:\TOOLS\NETTOOLS\SNMP (assuming E: is your CD-ROM)

Once installed, any SNMP Manager will be able to set traps and get information, based on the SNMP version 1 specification.

Accessing UNIX Hosts with Windows 98

As noted above with the TELNET utility, accessing a UNIX host is made easy from Windows 98. Run the command line TELNET to start the GUI interface, or create a shortcut on your desktop to it (right-click desktop/New/Shortcut, enter TELNET, press Enter twice), then click the shortcut. If you know the address or the name of the host you can start the TELNET program with this host name as the only argument on the line (or update your shortcut with this host name) as in:

TELNET UNIXHOST

Once you are connected to the UNIXHOST machine, you will be prompted to log in. This login is requesting you to enter a valid login name from the UNIXHOST passwd file, nothing to do with the Windows login names although they may match. You will also be prompted for a password for that login name if one is needed. After a successful login, your process will be started and you will be in some program, usually a program called a shell that is similar to a DOS command line, although, again, you could be starting a program designated by the UNIXHOST administrator for that particular login. To exit the command line shell you would type in EXIT and press return. This would end your login session on the UNIXHOST machine and return you to a blank TELNET program on your Windows 98 machine.

Table 25.3 Basic DOS and UNIX Command Comparison with Examples

DOS UNIX
DIR D:\wfw\system ls /c/wfw/system
COPY C:\config.sys c:\bck cp /c/config.sys /bck
REN C:\config.old config.new mv /c/config.old /c/config.new
DEL C:\config.001 rm /c/config.001
DELTREE \app1 \app2 rm -r /app1 /app2
TYPE config.sys cat /c/config.sys
CD \wfw\system cd /c/wfw/system
A: (N/A)

that UNIX uses the forward slash whereas DOS uses the backslash. There is a UNIX equivalent for almost every DOS command. Consult a UNIX reference for more on UNIX commmands.

Personal Web Server Service

An optional package contained on the installation CD, within the \ADD-ONS\PWS directory is the Personal Web Server installation package; use PWSSETUP.EXE to install.

E:\ADD-ONS\PWS\PWSSETUP.EXE  (if E: is your CD-ROM)

This service makes your Windows 98 system into a "mini" web page server. This is designed for light usage and is not intended as a full Web server service, this is a weak cousin of Internet Information Server, IIS, for NT.

You should install the Personal Web Server service only if you want others to access your web published documents from your machine directly instead of from some central server. If no central server exists, this is an easy way to provide basic local web services. Access to your machine will be via Internet Explorer or any other web browser.

that you must create the actual web pages with any text editor or any advanced web publishing tool like FrontPage 98 from Microsoft. You'll learn more about Web publsihing and Personal Web Server in Chapter 29, "Windows 98 as an Internet/Intranet Server."

Common Problems and Solutions

Some of the most common problems come from bad configuration or incomplete configuration of your IP information. Minimally you need at least a unique IP address and a network mask. Optionally you need a Default Gateway IP address if there are other networks to reach. Other useful options are DNS and/or WINS server IP addresses for host name resolution and improved browsing functionality (from WINS only). If you are still having some problems, check below for some common solutions.

No network:

Network card is not functioning properly, incorrect setup parameter. Go back to the diagnostics that came with the NIC and test it. Most NIC's provide a diagnostic and setup utility. You need two machines with the same type of NIC to run the network test. One machine must be set to send packets, the other is set to receive. If you have many NIC's to set up and test, you can use the same host to send packets to all other hosts as receivers. This will provide standard connection and diagnostic tests.

Cannot see any other machines in network neighborhood:

Wrong subnet or incorrect IP address for this network, check your settings.

Wrong mask for this network or subnet, check your settings in Network applet of Control Panel, click on the TCP/IP protocol, and then click Properties.

NIC not functioning correctly, see No Network: above.

Cabling not correctly installed or connected for network. Always check your cabling to ensure it works with at least one host. Test all cables against this one host to ensure they work.

Cannot Locate Another Host (with PING, TELNET, FTP, etc)

Wrong subnet or incorrect IP address for this network or

Wrong mask for this network or subnet, see notes above.

NIC not functioning correctly, see No Network: above.

Cabling not correctly installed or connected for network.

Try pinging your software, then your address or another host, first host may be down.

Duplicate IP

You have configured the same IP as another machine OR DHCP is leasing out an IP that another machine has already taken (Service Pack 3 on NT updates the DHCP server so that it PINGS the next available lease address three times by default on the network before allowing lease, thus ensuring no duplicate addresses are used). Just try another one or go talk to someone who sets the IP addresses. You can always PING the host from another running machine to see what IP it is then check with WINS or DNS for who should have it. The PING will return a MAC address, so you can always track that down as well by actually going to every machine on the network and checking the MAC address on each NIC.

Cannot Browse the rest of the network

Wrong IP address or network mask, check your settings against network settings.

WINS may not be configured properly or at all, need WINS to browse past your LAN.

Wrong Default Mask used, not getting to the remote network.

Wrong Gateway address used, not getting to the remote network, check settings in all cases.

Slow Response to PING, TELNET requests to other hosts

WINS and/or DNS services incorrectly set up, requests must time out first before you can continue. Check whether the IP addresses are correct for either or both of these services.

Telnet or FTP session very slow to start or never completes connection

Host name is not resolving back to you correctly. The remote host may need an entry in a host file for them to resolve back to you. Press Control+C to stop the program at the command prompt. You may have to wait up to 2 minutes for the GUI interface to respond when there is a network error but it should eventually time-out, be patient or do something else in another window.

Cannot `See' This Machine with SNMP

Intall or re-install service, check if service is running correctly in Control Panel Services.

Check that your network is set up correctly as indicated above, see No Network.

Cannot Run X Windows from UNIX

You need to install a third-party X Server product to run an X session on Windows 98. There are many available like www.Hummingbird.com and www.Intergraph.com.

Conclusion

This chapter described the TCP/IP protocol and you can configure and manage in Windows 98. You learned about IP addressing and name resolution and about some of the TCP/IP utilities, such as TELNET, FTP, PING, WINIPCFG, and more. This chapter also discussed SNMP and described some of the pitfalls you may encounter when setting up or using TCP/IP in Windows 98.

Previous chapterNext chapterContents

� Copyright, Macmillan Computer Publishing. All rights reserved.