8.4 Constrained Distribution of Routing Information

Recall that we defined a VPN by a set of policies that determine, among other things, connectivity among sites. Therefore, one of the mechanisms we need is a mechanism that can control connectivity among sites. The mechanism we use to accomplish this is constrained distribution of routing information. The reason why constrained distribution of routing information provides control over connectivity is because the flow of data is determined by the flow of routing information; therefore, constraining the flow of routing information constrains the flow of data. ^[3] Another way to put it is to say that connectivity and the flow of data are controlled by routing tables on the routers, and the content of those routing tables can be controlled by constraining the flow of routing information.

To understand how constrained distribution of routing information is used in the context of the BGP/MPLS VPN approach, let s first look at the distribution of routing information, which can be decomposed into five steps:

1. The routing information is propagated from a customer site to the service provider. More precisely, the information is propagated from a CE router to the PE router to which the CE router is connected. There are several options for propagating this information, such as RIP, static routing, OSPF, or BGP.

2. At the ingress PE router, this information is exported into the provider s BGP.

3. This information is distributed within the service provider among the PE routers using BGP.

4. This step is exactly the...