Introduction

Open Shortest Path First (OSPF) was developed as an open, non–proprietary protocol for TCP/IP. It is currently the most common routing protocol in use in enterprise and enterprise data centre networks. It was standardized (v2) for IPv4 as RFC2328 and for IPv6 as RFC5340.

Compared to RIP;

• There are no limitations on route costs

• IP multi-cast is used to send updates

• Updates are only sent after topology changes

• OSPF is efficient, it only sends updates after a topology change

• OSPF converges more quickly than RIP

• OSPF allows for the division of a network into areas and along with route summarization, this reduces the size of routing tables and makes the protocol scalable

• OSPF is classless

Using an IP address on an interface would not be a good way to identify a router. In OSPF, each router has a unique 32-bit router ID.

OSPF is a link state protocol.

A link is an interface on a router.

The state of that link are its properties and its relation to other routers.

When routers are connected and have OSPF configured they exchange OSPF Hello Packets and become neighbours. Each router sends Link States Advertisements or LSAs to these neighbouring routers, describing all its links and connections, all its interfaces. A collection of such information constitutes a Link-State Database or LSDB which each router holds, with a topology picture of the network. A protocol of this kind creates a map of the entire network, known as a graph. Next the router runs a Shortest Path First (SPF) algorithm which calculates the cheapest route to every node and subnet in the network. Finally, these optimal routes are loaded into the routing table.

When OSPF was first standardized in 1989, processors were not very powerful and did not have significant memory compared to today. Calculating SPF for a very large network would have been infeasible. The solution was to break an OSPF network up into areas and to calculate shortest path routes only within each area. A large OSPF network is made up of logical areas, with a central backbone area which is always numbered as zero.

The area number is a 32-bit value and many administrators use the dotted decimal notation of IP to represent areas; area zero will be written as 0.0.0.0 and area ten written as 0.0.0.10. A router with an interface in area 0.0.0.0 and another area is called an Area Border Router or ABR. The ABR summarises the routes in its area, keeping both the algorithm and routing table simple. Traffic between areas must travel via the backbone area 0.0.0.0, you can’t connect area 0.0.0.10 directly to area 0.0.0.20.

OSPF can take routing information from other routing protocols; we call this injecting routes or redistribution. The router which does this at the border of an OSPF network is an autonomous system border router or ASBR.

A router can have connected interfaces, static routes and routes from various dynamic protocols. Administrative distance defines the preference for a source of routing information. OSPF has a default administrative distance of 110, preferable to RIP (120) and internal BGP (200).