Unoverica Message Transport v2.11 Appendix B: TCP/IP-In-Depth

Appendix B: TCP/IP-In-Depth

B.4 IP Routing

IP addresses represent two pieces of information: part of the address indicates the IP network that the device is on, while the remainder of the address indicates the specific address of the specific device on that network. For example, 192.168.1.10 represents node "10" on network "192.168.1.0".

By using discrete network and node addresses, IP can route packets across multiple networks based on the network portion of the address. This allows the routers to aggregate routes for groups of systems, as opposed to having to store routing information for the theoretically possible 4.7 billion possible nodes. Aggregating allows the routers to only store routing information for a few million networks. While this is still too high for most systems to handle, it is much better than 5 billion. Once a packet reaches the destination network, it is routed to the specific node specified by the device’s address.

This is conceptually similar to the way that area codes and prefixes help the phone company route a call. The telephone switches can route a call to a remote area by examining the area code, at which point a local switch can route the call to a specific part of town based on the telephone number’s local prefix. Another switch (such as an internal PBX) can then route the call to a specific location (such as a specific office) by examining the last four digits of the phone number. This hierarchical routing mechanism allows for a distributed delivery approach which is easier to manage than a centralized database would allow for.

However, unlike the telephone network, IP networks are not necessarily aggregated according to geography: network 192.x.x could be located anywhere. Because of this, routers must have the ability to learn about the network. There are several common methods for this, ranging from manually configuring the router to forward all non-local packets to a predefined system, to high-end abstraction protocols that attempt to aggregate networks together based on portions of the network number.

There are three common types of routing services in use on Novell networks:

Static Routing. By manually defining the routers to be used for remote networks, systems do not have to use any advanced routing protocols. For more information on this subject, refer to section 1 Static Routing.
Routing Information Protocol (RIP). Routers and end-systems that support the RIP protocol can monitor the network for routing changes, automatically adjusting their maps to reflect the current network topology. RIP is effective for small networks, but does not scale well for larger environments. For more information on RIP routing, refer to section 2 Routing Information Protocol (RIP).
Open Shortest Path First (OSPF). Routers and end-systems that support the OSPF protocol can also monitor the network for routing changes, similar to RIP-enabled systems. However, OSPF is much more robust (and complex) than RIP, and is most appropriate for large-scale, complex networks. For more information on OSPF, refer to section 3 Open Shortest Path First (OSPF).