Cloud Native Data Center Networking

Chapter 13. Deploying OSPF

It was 20-minute invention...I designed it without pencil and paper. Without pencil and paper you’re forced to avoid all avoidable complexities.

Edsger Djikstra

This chapter begins the actual deployment details in a Clos topology: here, the link-state rubber meets the OSPF road. We explore configuration of the OSPF routing protocol for use in a Clos topology. I picked OSPF because within the data center, it is more popular than its older and better-designed cousin, IS-IS. Both protocols, however, use the shortest path algorithm—invented by the Dutch computer scientist, Edsger Djikstra—to build the routing table. As he states in the opening quote, it is a simple and elegant algorithm and the cause of much of his fame.

The goal of this chapter is to help network engineers determine the ideal OSPF configuration for their networks. This chapter should help answer the following questions:

• When is OSPF useful in the data center?

• What are the key design principles for configuring OSPF?

• What are the differences between OSPFv2 and OSPFv3, and how should I use them?

• How do I configure OSPF in my routing stack?

• How do I configure OSPF on servers, for example, to provide a routed environment for containers?

• How do I use OSPF to help upgrade the router software?

We begin by describing the problems we need to solve. We then map these problems into OSPF concepts that address them, before finally describing the actual configuration. If you’re already ...