In the real world, computer networks can combine a variety of physical and logical topologies. For the sake of simplicity and clarity, all of the subnetworks in the following discussion will be based on the NetWare client-server networking model.
So far we have discussed basic internetworks existing at one local site, with computers and other devices directly connected by some type of cabling. These types of networks are commonly referred to as LANs. In real-world applications, however, LANs are much more complex than the simple models we have discussed. In addition, there are networks composed of complex multiserver internetworks that exist at separate sites—which might be any number of miles apart—connected by long-range transmission media. These types of networks are known as WANs. The following sections discuss internetwork configurations that are more likely to be encountered in real-world situations.
One Server, Multiple Networks of the Same Type
The simplest form of an internetwork is two cabling (media) segments of the same MAC type sharing one network server.
For example, one server could contain two Ethernet network adapters, each supporting a different cable segment. There could be several computers connected to each cable segment in a star physical layout, with each cable segment using contention (CSMA/CD) for the MAC. Each of the cable segments would have a different network address; thus, each would be an independent subnetwork.
Together, the two separate networks would form an internetwork connected by internal routing capabilities built into the server.
Figure 32 illustrates the one-server internetwork described above.
Figure 32: Internetworking two networks using the same type of network adapter in one NetWare server, by means of the server's internal routers
In the network illustrated in Figure 32, routing would be performed using NetWare IP or NetWare IPX, with support from the other NetWare routing protocols, as previously described.
Every NetWare server is capable of using internal routers to accomplish local network routing by means of the NetWare routing protocol set and AppleTalk. For larger or more complicated internetworks, or for departments with heavy server-processing requirements, NetWare or dedicated routers from other vendors provide the necessary routing power and capabilities.
One Server, Multiple Networks of Differing Types
In a slightly more complex internetwork, a NetWare server could support multiple cable segments using the same physical layouts but different MACs.
For example, a server could contain one Ethernet network adapter and one token-ring network adapter, with a cable segment attached to each. The Ethernet network might be connected in a physical star and use CSMA/CD for the MAC. The token-ring network might also be connected in a physical star, but it would use token passing for the MAC. Like the simpler configuration explained in the previous section, each cable segment would have a different network address. Figure 33 illustrates this more complex one-server internetwork.
Figure 33: Internetworking two networks using different types of network adapters in one NetWare server, by means of the server's internal routers
In the case of the internetwork shown above, routing would again be performed using NetWare IP or NetWare IPX, with support from the other NetWare routing protocols.
Both one-server networks illustrated above support only two separate subnetworks. However, all NetWare servers are capable of supporting as many as 32 different network adapters (32 separate subnetworks) in any combination of same or different types.
Even though the token-ring network above was described as a physical star, it is drawn as a ring to signify that it is a token-ring network. In nearly all illustrations it is more important to make the logical topology clear than to be concerned with the physical topology.
Multiple Servers with Multiple Networks of Different Types
In an even more complex internetwork there may be multiple servers. For example, a complex internetwork might consist of two one-server subnetworks connected by a standalone router. Each server might contain multiple network interface adapters.
One server might contain two Ethernet network adapters and one token-ring network adapter, with a cable segment attached to each. One of the Ethernet adapters might support a PC network, and the other might connect to both PCs and Macintosh computers.
The other server might contain two Ethernet adapters that support PCs and Macintoshes. On one Ethernet, some nodes might be APs supporting wireless PCs and handheld devices.
Each server would have a unique internal number (server address), and each cable segment in each server would have a unique physical network (cable segment) address.
In this case, there would be five subnetworks on the internetwork: three attached to one server and two attached to the other. The internal server routers would perform the routing between any two workstations on subnetworks attached to the same server. Both the internal server routers and the intermediate standalone router would be involved in the routing between any two workstations on subnetworks attached to different servers.
Figure 34 illustrates the two-server internetwork described above.
Figure 34: Internetworking multiple networks using different types of network adapters in two NetWare servers, by means of internal and standalone routers
An already complex multiserver internetwork becomes even more complex with the addition of connections to host computer systems: mainframe computers such as IBM mainframes, minicomputers such as IBM's AS/400 or a VAX system, or other hosts such as UNIX workstations.
Host systems can provide access to other application software, additional resources such as data storage devices and printers, and increased processing power. For example, you might want to log on to an IBM AS/400 minicomputer to run an application available only on that computer, or to use its processing power for one task while you were using the processing power of your own workstation for another task. Or, you might want to print a large report on a high-speed printer connected to an AS/400 minicomputer.
The illustration in Figure 35 shows a multiserver NetWare network with an IBM mainframe, an IBM AS/400 minicomputer, and several UNIX workstations connected as host computers.
Figure 35: Host systems connected to a complex multiserver NetWare network
As a recent addition to the world of computer networking, clustering services take networking to a new level. With the services, you can combine several servers on your network into a single unit with increased fault tolerance. The servers are "clustered" using specific software and programmed to keep tabs on one another so that the failure of any individual server will not bring the network down. When one server goes down, the others can immediately step in and provide the services it offered. This allows the administrators to repair downed servers or perform routine maintenance without interrupting network productivity.
Wide Area Networking
The traditional definition of wide area networking has been "connecting two or more networks existing at widely separate geographic sites." Some traditionalists also specify that the separate networks must be connected by means of common carrier telecommunication facilities (private companies that rent resources such as T1 lines and microwave transmission equipment).
Like any general term used in connection with rapidly changing technology, not everyone will agree on an exact definition of wide area networking. What is "widely separate"? Does the connection really have to be through a common carrier? Many major companies now own their own equipment linking networks that are many miles apart.
Suppose you connect two networks in two different buildings 100 yards apart by means of asynchronous modems and common telephone lines. Is that wide area networking? Most networking experts would say no: they would describe it as "one-site" or "campus" networking. What if the networks were two miles apart and separated by a major interstate highway? Or, what if they were 15 miles apart, on opposite sides of a major city? Many experts would still not call this wide area networking; they might use the term "metropolitan area networking." Others consider metropolitan area networking a part of wide area networking.
Figure 36 shows two separate branch office internetworks connected to a third internetwork at a main corporate office. Each internetwork has multiple servers and existing host connections. One of the branch office networks is connected to the corporate network with a wireless router. The other branch office network is connected with a common carrier-provided intermediate link: in this case, a frame relay packet-switching network. However, either network could be connected by other means such as ATM or a dedicated leased-line link, perhaps using PPP.
Figure 36: Wide area networking: three networks at widely separated sites connected through wireless routers and a frame relay connection
Return to Primer Index | Next Section