[isp-bgp] BGP Loop Detection Mechanism doubt.

["Murilo Antonio Pugliese" <mpugliese@...>]

Hi folks.

As my first message to this list, I may introduce myself.
My name is Murilo Pugliese, I'm an electrical engineer, and work for a telecommunications company at Brazil that among 
other services (as those one related to IDCs) provide Internet access for companies, MPLS-based VPN, and VoIP services. 
I have one doubt regarding the BGP Loop Detection that I'm expecting to get it clear through this list.

Talking about MPLS-VPN deployments, I have a scenario (common request) were a VPN customer has a site with 2 Wan 
connections & wants that this site receive its downstream traffic load-shared. 
At first we just associated both wan connections to the same VRF, issued the load-sharing command at interfaces level, 
and created parallel routes toward the site LAN prefix (each one pointing to one of its wan connection).

We could perceive that this deployment worked just fine for traffic originated at VRFs present @ different/remote PEs;
but for traffic originated at a VRF present @ the same PE the source VRF would choose only one adjacency and send all 
its traffic just through this chose adjacency, so the downstream load-sharing request wasn't being provide as required.
 
To overcome this for sites with this kind of need, we decided to established eBGP between PE and CE, though the devices 
loopback addresses. In this way, doesn't matter if traffic is originated from a VRF at the same PE or from a remote PE, the 
downstream traffic is always load-shared.

Now comes my doubt. If couples sites of the same VPN are configure using this solution, and both sites establish BGP using the 
same AS #, when one VRF gets to import a prefix propagated through my MPLS Domain from the remote VRF, it's possible that
the BGP Loop Detection Mechanism interpret it as a loop and discard the prefix ? 

Be advised that there is no BGP establishment between both VRFs, only between each VRF and its peer CE.

The command below was issued at a VRF that imports a prefix from a remote VRF that uses AS # 6500 to establish eBGP with its 
peer CE. I know there is no AS Path, but there is a reference of the AS # associated to the routing information.

	Router#sh ip cef vrf 55123456789 0.0.0.0 0.0.0.0
	0.0.0.0/0, version 757, epoch 0, per-destination sharing
	0 packets, 0 bytes
	Flow: AS 65000, mask 0
	tag information set, all rewrites owned
	local tag: VPN route head
	fast tag rewrite with
	Recursive rewrite via A.B.C.D/32, tags imposed {944}
	via A.B.C.D, 0 dependencies, recursive
	next hop A.B.X.Y, FastEthernet5/0/0 via A.B.C.D/32 (Default)
	valid adjacency
	tag rewrite with
	Recursive rewrite via A.B.C.D/32, tags imposed {944}
	Recursive load sharing using A.B.C.D/32.
	Router#

Willing to hear from you folks ASAP.
Yours Truly.
Murilo Pugliese.






To unsubscribe via postal mail, please contact us at:
Jupitermedia Corp.
Attn: Discussion List Management
475 Park Avenue South
New York, NY 10016

Please include the email address which you have been contacted with.

Copyright 2005 Jupitermedia Corporation All Rights Reserved.