UNIX System Administration Handbook - Evi Nemeth [211]
Although RIP is a resource hog because of its profligate use of broadcasting, it does a good job when a network is changing often or when the topology of remote networks is not known. However, it can be slow to stabilize after a link goes down.
RIP is widely implemented on non-UNIX platforms. A variety of common devices from printers to SNMP-manageable network components can listen to RIP advertisements to learn about possible gateways. In addition, almost all versions of UNIX provide routed, so RIP is a de facto lowest common denominator routing protocol. Often, RIP is used for LAN routing and a more featureful protocol is used for wide-area connectivity.
RIP-2: Routing Information Protocol, version 2
RIP-2 is a mild revision of RIP that adds support for a few features that were missing from the original protocol. The most important change is that RIP-2 distributes netmasks along with next-hop addresses, so it provides better support for subnetted networks and CIDR. A vague gesture towards increasing the security of RIP was also included, but the definition of a specific authentication system has been left for future development.
See page 276 for information about classless addressing, aka CIDR.
Many sites use routed in its -q (“quiet”) mode, in which it manages the routing table and listens for routing updates on the network but does not broadcast any information of its own. At these sites, the actual route computations are usually performed with a more efficient protocol such as OSPF (see below). The computed routes are converted to RIP updates for consumption by nonrouter machines. routed is lightweight (in -q mode) and universally supported, so most machines can enjoy the benefits of dynamic routing without any special configuration.
RIP-2 provides several features that seem targeted for this multiprotocol environment. “Next hop” updates allow broadcasters to advertise routes for which they are not the actual gateway, and “route tags” allow externally discovered routes to be propagated through RIP.
RIP-2 can be run in a compatibility mode that preserves most of the new features of RIP-2 without entirely abandoning vanilla RIP receivers. In most respects, RIP-2 is identical to RIP and should be preferred over RIP if it is supported by the systems you are using.
OSPF: Open Shortest Path First
OSPF is defined in RFC2328. It’s a link-state protocol. “Shortest path first” refers to the mathematical algorithm used to calculate routes; “open” is used in the sense of “nonproprietary.”
OSPF was the first link-state routing protocol to be broadly used, and it is still the most popular. Its widespread adoption was spurred in large part by its support in gated, a popular multiprotocol routing daemon of which we will have more to say later. Unfortunately, the protocol itself is very complex and hence only worthwhile at sites of significant size (where routing protocol behavior really makes a difference).
The OSPF protocol specification does not mandate any particular cost metric. gated’s implementation uses hop counts by default, as does Cisco’s. Cisco routers can also be configured to use network bandwidth as a cost metric.
OSPF is an industrial-strength protocol that works well for large, complicated topologies. It offers several advantages over RIP, including the ability to manage several paths to a single destination and the ability to partition the network into sections (“areas”) that share only high-level routing information.
IGRP and EIGRP: Interior Gateway Routing Protocol
IGRP and its souped-up successor EIGRP are proprietary routing protocols that run only on Cisco routers. IGRP was created to address some of the shortcomings of RIP before robust