1

lately I have been checking out some old IOS-XE PEs in the environment that are peering with each other directly before route reflectors were introduced.

they are using BGP autodiscovery and signaling. surprisingly I found in the config that they are not sending extended community with the 'send-community extended' command but works just fine.

this makes me wonder if route target and VPLS ID are only needed when the PEs need to find each other through route reflectors.

if so, could it also be the case for L3VPN and EVPN? though I think layer 3 CE routes redistribution needs route target regardless.

Improve this question

2 Answers 2

Reset to default
1

Route targets don't help PE routers find each other. And PE routers actually don't need to find each other, it's the BGP service they're running that has to find its peer to establish a BGP session. The sole function of a route target is to control which VRF/EVI routes the sending PE should export, and which routes the receiving PE should import into a VRF/EVI.

Improve this answer
1
  • I should rephrase it: RT and VPLS ID help identify which PEs belong to the same VPLS domain. however, they are not needed if BGP are directly peered?
    – daniel gor
    Commented Apr 28 at 5:57
0

just tested the same setup with JUNOS. as soon as I mismatched either VPLS ID or route target the service went down.

no wonder why so many say VPLS is the least standardized when it comes to implementation.

ps: also tested EVPN in IOS XE. the service goes down in the next BGP advertisement after I removed 'send-community extended' command.

Improve this answer

Not the answer you're looking for? Browse other questions tagged or ask your own question.