Route distinguisher

From Hill2dot0
(Redirected from RD)
Jump to: navigation, search

Many students new to MPLS are confused about the difference between the route distinguisher and the route target. This is understandable since both share a similar format and, depending on the MPLS platform being configured, are displayed near each other. However, the two are very different in function. The paragraphs that follow explain the need for the route distinguisher and provide more background about its use.

One of the major benefits of the MPLS/BGP VPN is that it allows the customer to use any address space it desires. In a large service provider network, it is quite likely that customers are using the same address space.

Route distinguisher

As the visual indicates, customer VPN ABC is using the same address space as customer VPN XYZ. The PE router for each receives the IP route from the customer and both advertise it to the IBGP peer PE-1. Since PE-1 is only running a single BGP process, from its point of view, it has received two IP routes for the same network. Under the normal rules of BGP, only one of these routes will be accepted as the “best” route and placed into the IP routing table. This is clearly unacceptable, since the two route updates represent two different customer VPNs. Depending on which route the BGP process on PE-1 chooses, one VPN will disappear from the routing table, effectively disconnecting it from the network.

To solve this dilemma, the PE-1 router needs a way to discern that the two route updates it has received are part of different VPNs and should be treated as two different routes. One way to do this is to run multiple BGP processes for each VPN. Of course, this would get complicated quickly. The other alternative is to uniquely identify each route. This is accomplished by prepending a route distinguisher to all routes advertised by the customer PEs.

Since a single VPN cannot have overlapping address space, the route distinguisher can be assigned on a per VPN basis, with all sites sharing the same route distinguisher. However, this method might cause problems when implementing complex VPN topologies where a single site becomes part of multiple VPNs. Typically, a single site is assigned a single route distinguisher. Since a single site has its own VRF, this also means that there is typically a one-to-one relationship between VRFs and route distinguishers.