Routers

Routers have been known by several names . Back in ancient times when what is now the Internet was called the ARPANET, routers were called IMPs, for interface message processors . ^[8] More recently, routers were called gateways ; remnants of this nomenclature can still be found in terms such as Border Gateway Protocol (BGP) and Interior Gateway Routing Protocol (IGRP). ^[9] In the Open System Interconnection (OSI) world, routers are known as Intermediate Systems (IS).

^[8] The parent of modern packet-switched networks was the AlohaNet, created at the University of Hawaii in the late 1960s by Norman Abramson. Because routers at that time were called IMPs, Dr. Abramson rather impishly named his router Menehune: a Hawaiian elf.

^[9] The term gateway is now generally accepted to mean an application gateway, as opposed to a router, which would be a network gateway.

All of these aliases are descriptive of some aspect of what a router does. As interface message processor implies, a router switches data messages, or packets, from one network to another. As gateway implies, a router is a gateway through which data can be sent to reach another network. And as Intermediate System implies, a router is an intermediary for the End System “to “End System delivery of internetwork data.

Note

Router

Router , as a name , is probably the most descriptive of what the modern versions of these devices do. A router sends information along a route ”a path ”between two networks. This path may traverse a single router or many routers. Furthermore, in internetworks that have multiple paths to the same destination, modern routers use a set of procedures to determine and use the best route. Should that route become less than optimal or entirely unusable, the router selects the next -best path. The procedures used by the router to determine and select the best route and to share information about network reachability and status with other routers are referred to collectively as a routing protocol .

Note

Routing protocol

Just as a data link may directly connect two devices, a router also creates a connection between two devices. The difference is that, as Figure 1.8 shows, whereas the communication path between two devices sharing a common data link is a physical path, the communication path provided by routers between two devices on different networks is a higher-level, logical path.

Figure 1.8. A router creates a logical path between networks.

Note

Packet

This concept is vitally important for understanding a router's function. Notice that the logical path, or route, between the devices in Figure 1.8 traverses several types of data links: an Ethernet, an FDDI ring, a serial link, and a Token Ring. As noted earlier, to be delivered on the physical path of a data link, data must be encapsulated within a frame, a sort of digital envelope. Likewise, to be delivered across the logical path of a routed internetwork, data must also be encapsulated; the digital envelope used by routers is a packet .

As noted earlier, each type of data link has its own unique frame format. The internetwork route depicted in Figure 1.8 crosses several data links, but the packet remains the same from end to end.

How is this possible? Figure 1.9 shows how the packet is actually delivered across the route:

1. The originating host encapsulates the data to be delivered within a packet. The packet must then be delivered across the host's data link to the local router ”that host's default gateway ”so the host encapsulates the packet within a frame. This operation is the same as placing an envelope inside of a larger envelope, for example, inserting an envelope containing a letter into a Federal Express envelope. The destination data link identifier of the frame is the identifier of the interface of the local router, ^[10] and the source data link identifier is the host's.

   ^[10] Although the purpose of a router is to create pathways between data links (networks), the router must also obey the protocols of the networks to which it is attached. So a router interface connected to an Ethernet will have a MAC identifier and must obey the CSMA/CD rules, a Token Ring interface must obey Token Ring rules, and so forth. In other words, a router is not only a router, but also a station on each of its attached networks.

2. That router (router A in Figure 1.9) removes the packet from the Ethernet frame; router A knows that the next-hop router on the path is router B, out its FDDI interface, so router A encapsulates the packet in an FDDI frame. Now the destination identifier in the frame is the FDDI interface of router B, and the source identifier is the FDDI interface of router A.

3. Router B removes the packet from the FDDI frame, knows that the next-hop router on the path is router C across the serial link, and sends the packet to C encapsulated in the proper frame for the serial link.

4. Router C removes the packet and recognizes that the station for which the packet is destined is on its directly connected Token Ring network; C encapsulates the packet in a Token Ring frame with the destination identifier of the destination station and the source identifier of its Token Ring interface. The packet has been delivered.

Figure 1.9. The frame changes from data link to data link, but the packet remains the same across the entire logical path.

The key to understanding this entire process is to notice that the frames and their related data link identifiers, which have relevance only for each individual network, change for each network the packet traverses. The packet remains the same from end to end.

But how did the originating host know that the packet needed to be delivered to its default gateway for routing? And how did the routers know where to send the packet?