MPLS VPN

[NaungHtunLwin0780]

How VRF's CEF know transport label and next hop that do not associate with this VRF . Isn't true that VRF control plane and  global control plane is separated . Please someone help me to solve my confusion about it . 

[balaji.bandi]

here is the good presenation which can help you to understand better with some example :

https://www.ciscolive.com/c/dam/r/ciscolive/us/docs/2017/pdf/BRKCCIE-3345.pdf

[NaungHtunLwin0780]

Thank you for your suggestion .

My confusion is just in here ! ( attached in photo )

In this presentation , he use " show ip cef 4.4.4.4 " that is checking global cef FIB .

Shouldn't he use " show ip cef vrf <vrf name> 4.4.4.4 " to check VRF FIB .

[Giuseppe Larosa]

Hello @NaungHtunLwin0780 ,

in MP BGP address family vpnv4 the BGP next-hop is actually the loopback address of the remote PE node and it is an address belonging to global routing table.

For this reason the slide checks with the commands

show mpls forwarding 4.4.4.4

and

show ip cef 4.4.4.4

The 4.4.4.4 is the address of the remote PE loopback that is the destination of the LSP created by LDP and the external label (also called IGP label) is 202.

The second inner label is provided by MP BGP and it is allocated from the label space of remote PE node and uniquely identifies the VRF and prefix on the remote PE node.

This label travels unchanged along the path between source PE and destination PE.

The external label changes at each router hop and follows the LSP towards the remote PE loopback 4.4.4.4/32 in this case.

This is a example of BGP recursion where the VPNv4 prefix is resolved over an IPv4 next-hop that is in GRT.

The BGP next-hop provides a pointer to the correct MPLS LSP built by LDP to be used to send traffic in the VPN with two MPLS labels (label stack).

Hope to help

Giuseppe

[NaungHtunLwin0780]

Thank you for your explanation . I understand what you mean and your answer is perfect for LSP process . Just in my mind , I actually want to know the detail process of CEF in MPLS L3 VPN .

[Giuseppe Larosa · Accepted solution]

Hello @NaungHtunLwin0780 ,

CEF is a pre-requisite for MPLS and for MPLS L3 VPN.

On the local PE node the MP BGP allocates a free MPLS label to each IPv4 prefix learned from the CE on the locally attached VRF on the VRF access link by any means (static, BGP or other routing protocol that requires redistribution into BGP address-family ipv4 vrf <vrf-name>).

The per VRF CEF tables are organized as the main CEF tables with an adjacency table that keeps the information for the next-hops ( CE IP addresses and their encapsulation including per VRF ARP entries for LAN interfaces) and there is a FIB that contains all the locally learned prefixes and all the prefixes resulting from the import from remote PE nodes.

In the per node LFIB there is one entry for each VPNv4 prefix.

When the local node reiceves a frame with only the VPN label ( for the PHP penultimate Hop Popping removing the external label) it uses the label value to find the action to perform. This action is indicated with a capitol A aggregated and becomes a pointer to the per VRF CEF entry of the prefix allowing to send the packet out the correct interface with the correct encapsulation in order to be received and processed by the CE node as a standard IPv4 packet.

Hope to help

Giuseppe

[NaungHtunLwin0780]

Thank you very much . I accept as the answer of my confusion . I appreciate your patient and kindness . thank you ver much .

[balaji.bandi]

Nice @Giuseppe Larosa  covered most stuff here. no extra word to say here.