Chapter 12. Link-State Protocols: Open Shortest Path First (OSPF) Since its conception in 1987, OSPF has continued to evolve with the modern internetwork. Today, OSPF is clearly the dominant Open Systems interior routing protocol. Because of the constant RFC enhancements to OSPF, it has proven itself to be a scalable protocol with networks in the tens of thousands of nodes. OSPF officially was documented in 1989 by the Internet Engineering Task Force (IETF) to address many of the limitations of distance vector protocols, such as RIP and IGRP. Since 1989, OSPF evolved through several RFCs, beginning with Version 1 in RFC 1131 and moving to Version 2 in RFC 1247, which was superceded by RFC 2178, to its final form, Version II in RFC 2328. In modern large networks, OSPF become the standard open routing protocol. The word open is used because the algorithm that drives OSPF, Dijkstra's Shortest Path First (SPF) algorithm, isn't proprietary to any vendor or organization. This allows for mainframes such as IBM, Unisys, and DEC, as well as other manufacturers' routers, to run OSPF. As previously mentioned, OSPF provides many significant enhancements over current distance vector protocols: -
Fast convergence ” OSPF uses a reliable flooding mechanism to update neighboring routers of changes in network topology. Only partial routing updates are sent upon the loss of a route. These two facts, combined with the fact that all routers in the OSPF domain have a nearly identical view of the network, allow for OSPF to converge quickly. -
Support for VLSM, supernetting, and summarization ” OSPF can use summarization and VLSM to conserve address space and streamline routing. -
Support for large network diameters ” By using VLSM and forcing hierarchical network design based on areas, the OSPF network diameter is virtually limitless. -
Stub area routing ” Large networks are supported because stub-area routing reduces the routing table. -
Efficient and reliable transport of routing updates ” Through the use of two reserved multicast addresses to transport routing updates, OSPF does not impact non-OSPF routers and devices. OSPF updates are reliable because they all are implicitly or explicitly acknowledged . -
Efficient use of media ” Multicasts messages occur only on broadcast media. Unicast are used on NBMA and point-to-point networks. -
Arbitrary metric based on cost ” OSPF uses a metric of cost, which can be changed, to base routing decisions on. -
Equal-cost load balancing ” OSPF load-shares across equal-cost paths, optimizing bandwidth and multiple paths. -
Support for Type 1 and Type II (MD5) authentication ” OSPF allows for secure route exchange by using Type 1 clear-text passwords or Type 2 MD5 authentication. -
Support for route tagging of OSPF external routes ” Tags can be added to external routes redistributed into OSPF. Tags can be used as another control for routing policy within the autonomous system or for internal documentation. -
Fully classless routing protocol ” OSPF supports a classless route table lookup and is not susceptible to classful routing problems, such as discontiguous subnets. -
No susceptibility to split-horizon issues. In spite of these advantages, OSPF has been criticized as being overly complicated to configure and requiring high processor utilization. It is true that even the smallest OSPF network will require a small amount of design before implementation, and it does require more CPU cycles than other routing protocols. However, with the advent of modern CPUs, the amount of processor utilization for OSPF can be minimal. In this chapter, the technical operation of OSPF is discussed along with the configuration aspects of OSPF. |
