Automatic Route-Distinguishers vs. Manual Route-Distinguishers
Every MPLS VRF needs a unique route distinguisher. This is needed for BGP to tell the difference between two of the same prefixes on two different VPNs. You wouldn’t want the route-selection process to perform a route-selection between the two because they’re in separate routing domains.
Both JUNOS and IOS-XR have this functionality. I don’t think IOS has this functionality.
There really isn’t a better technical reason for using automatic distinguishers. The only reason I like them is for ease of setup and to erase the possibility of mis-typing distinguishers when setting up new VRFs. I’m a bit removed from operations, but I don’t remember caring too much about the distinguisher’s assigned-number field during issues.
Having manual distinguishers and knowing which manual distinguishers are where is valuable during troubleshooting.
Most networks typically use the VLAN-IDs of the interfaces in the VRF in the route-distinguisher. This is a good practice, however, the VLANs are usually not network-wide (typically locally relevant only) and are re-used throughout the network. So, seeing a route-distinguisher of 10.10.10.10:108 while in another router wouldn’t exactly tell you much about the route you’re looking at. It could actually cause confusion if the router you were on also had a VLAN 108.
Another way is to manually choose route-distinguishers and uniquely assign them starting at 1 going to 65535. In this case, you would refer to a spreadsheet/doc/app with the route-distinguisher documented to verify this information, when you could just as easily log into the PE and figure out the originating VRF.
We should start an RFC for ASCII RD fields 🙂
Which method does everyone else use and/or prefer?