Border Gateway Protocol (BGP) Policies in Routed Networks

• Simplify the BGP policy space into three basic kinds of policies that can be expressed
• Explore the many different methods available for configuring each of the three basic kinds of policies
• See practical examples that illustrate how to implement each policy

BGP is different from other routing protocols because it is designed to find loop-free paths using what might conventionally be called a “path cost” only after operator policies are considered. BGP, then, is focused on policies rather than metrics. For the network engineer, this shift in perspective from metrics to policies can be difficult to understand.

This course begins by simplifying the entire BGP policy space into three basic kinds of policies that operators implement using BGP—selecting the outbound path, selecting the inbound path, and “do not transit.” A use case is given for each of these three kinds, or classes, of policies from the perspective of a transit provider, and another from the perspective of a nontransit operator connected to the edge of the ‘net.

With this background in place, the course will then explore each of the many ways these classes of policy may be implemented using local preference, AS Path prepending, various communities, AS Path poisoning, and other techniques. Positive and negative aspects of each implementation path will be considered.