The terminology used for network addresses is different depending on the protocol in use. If the protocol in use is IPX, the logical network is simply referred to as a network address. With IP it is a subnet and with AppleTalk it is called a zone.
Note NetBIOS and NetBEUI are non-routable protocols, although NetBEUI can be thought of as overlapping the transport, network, and (the LLC portion of the) data-link layer. They do not use network numbers and do not have the ability to propagate information between logical network segments. A non-routable protocol is a set of communication rules that expects all systems to be connected locally. A non-routable protocol has no direct method of traveling between logical networks. A NetBIOS frame is incapable of crossing a router without some form of help.
Routers are used to connect logical networks, which is why they are sometimes referred to in the IP world as gateways. Figure 3.6 shows the effect of adding a router to a network. Notice that protocols on either side of the device must now use a unique logical network address. Information destined for a non-local system must be routed to the logical network on which the system resides. The act of traversing a router from one logical network to another is referred to as a hop. When a protocol hops a router, it must use a unique logical network address on both sides.
Figure 3.6: The effects of adding a router to the network
So how do systems on one logical network segment find out what other logical segments exist on the network? Routers can either be statically programmed with information describing the path to follow in order to find remote networks, or they can use a special type of maintenance frame such as the routing information protocol (RIP) to relay information about known networks. Routers use these frames and static entries to create a blueprint of the network known as a routing table.
Note Routing tables tell the router which logical networks are available to deliver information to and which routers are capable of forwarding information to that network.
Routing Tables
You can think of a routing table as being like a road map. A road map shows all the streets in a local city or town in much the same way a routing table keeps track of all the local networks.
Without having some method for each of these routers to communicate and let each other know who is connected where, communication between logical network segments would be impossible.
There are three methods for routing information from one network to another: Static
Distance vector Link state
While each protocol has its own ways of providing routing functionality, each implementation can be broken down into one of these three categories.
Static Routing
Static routing is the simplest method of getting information from one system to another. Used mostly in IP networks, a static route defines a specific router to be the point leading to a specific network. Static routing does
Internet), a single router may be designated as a default to receive all traffic destined for networks that have not been predefined. When static routing is used, most workstations receive an entry for the default router only. For example, let’s assume I configure my system to have a default route that points to the router Galifrey. As my system passes information through the network layer, it will analyze the logical network of the destination system. If the system is located on the same logical network, the data-link layer adds the MAC address of that system and transmits the frame onto the wire. If the system is located on some other logical network, the data-link layer will use the MAC address for Galifrey and transmit the frame to it. Galifrey would then be responsible for insuring that the frame gets to its final destination.
The benefits of this type of routing are simplicity and low overhead. My workstation is not required to know or care about what other logical networks may be available and how to get to them. It has only two possibilities to worry about—deliver locally or deliver to Galifrey. This can be useful when there is only one possible route to a final destination. For example, most organizations have only one Internet connection. Setting up a static route that points all IP traffic to the router that borders this connection may be the easiest way to insure that all frames are delivered properly. Because all my routing information is configured at startup, my routers do not need to share route information with other routers. Each system is only concerned with forwarding information to its next default route. I do not need to have any dynamic routing frames propagated through my network, because each router has been preset as to where it should forward information.
While static routing is easy to use, it does suffer from some major drawbacks that severely limit its application. When redundant paths are provided, or even when multiple routers are used on the same logical network, you may find it more effective to use a routing method that is capable of exchanging dynamic routing information.
Dynamic routing allows routing tables to be developed on the fly, which can compensate for hardware failures. Both distance vector and link state routing use dynamic routing information to insure routing tables stay up to date.