A router performs redistribution when it uses a routing protocol to advertise routes that were learned by some other means. Those “other means” may be another routing protocol, static routes, or a direct connection to the destination network. For example, a router may be running both an OSPF process and a RIP process. If the OSPF process is configured to advertise routes learned by the RIP process, it is said to be “redistributing RIP.”
Running a single routing protocol throughout the entire IP internetwork is usually more desirable than running multiple protocols, both from a configuration management perspective and from a fault management perspective. However, the realities of modern internetworking frequently force the acceptance of multiprotocol IP routing domains. As departments, divisions, and entire companies merge, their formerly autonomous internetworks must be consolidated.
In most cases, the Internet that are to be consolidated were implemented differently and have evolved differently, to meet different needs or merely as the result of different design philosophies. This diversity can make the migration to a single routing protocol a complex undertaking. In some cases, corporate politics can force the use of multiple routing protocols. And in a few cases, multiple routing protocols may be the result of network administrators who do not work and play well together. Multi-vendor environments are another factor that can necessitate redistribution. For example, an Internet running Cisco’s IGRP or EIGRP may be merged with an Internet using another manufacturer’s routers, which support only RIP or Open Shortest Path First. Without redistribution, either the Cisco routers would have to be reconfigured to an open protocol or the non-Cisco routers would have to be replaced with Cisco routers. Redistribution is necessary when multiple routing protocols are “thrown together,” but redistribution may also be part of a well-thought-out internetwork design. Figure 11.1 shows an example. Here two OSPF process domains are connected, but the OSPF processes do not directly communicate. Instead, static routes are configured on each router to selected networks in the other OSPF domain.
For instance, router Spalding has static routes to networks 192.168.11.0 and 192.168.12.0. Spalding then redistributes these static routes into OSPF, and OSPF advertises the routes to the other routers in OSPF 10. The result is that the other networks in OSPF 20 are hidden from OSPF 10. Redistribution has allowed the dynamic characteristics of OSPF to be mixed with the precise control of static routes.
Static routes redistributed into a dynamic routing protocol are also very useful, if not essential, in dial-up environments. The periodic management traffic of a dynamic protocol can cause the dial-up line to be “always up.” By blocking routing updates or hellos across the link and configuring static routes on each side, the administrator can ensure that the link will only come up when user traffic must traverse it. And by redistributing the static routes into the dynamic routing protocol, all routers on both sides of the dialup link have full knowledge of all networks on the opposite side of the link.
Note that with few exceptions , the existence of more than one routing protocol on the same router does not mean that redistribution will automatically occur. Redistribution must be explicitly configured. The configuration of multiple routing protocols on a single router without redistribution is called ships in the night (SIN) routing. The router will pass routes to its peers in each process domain, but the process domains will have no knowledge of each other—like ships passing in the night. Although SIN routing usually refers to multiple routing protocols routing multiple routed protocols on the same router (such as OSPF routing IP and NLSP routing IPX), it can also refer to two IP protocols routing for separate IP domains on a single router.