cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
785
Views
23
Helpful
5
Replies

Do the P routers running IGP in the SP CORE need to have all the CE + Other SPs prefixes in their routing table?

superkliko
Level 1
Level 1

Hi Experts,

I'm new in the SP world, could you please help me with the following.

In a network design see below, PEs are running iBGP, the PEs and Ps is running ISIS / MPLS.

My question is, does the P routers running ISIS need to have the full prefixes of the CE/other-SPs networks like the PE or just maintain the internal SP prefixes needed as a communication base for the MPLS/MBGP.

 

CE1 --> PE --> P1 -> P2 -> P3 -> P4 --> PE --> CE1

 

Thank you

 

 

1 Accepted Solution

Accepted Solutions

CE prfixes are known to only PE routers and its passed to other end PEs via iBGP (aka MP-BGP).

P routers only have IGP routes so that all PEs and Ps are accessable via each other (internal SP network rechability)

View solution in original post

5 Replies 5

CE prfixes are known to only PE routers and its passed to other end PEs via iBGP (aka MP-BGP).

P routers only have IGP routes so that all PEs and Ps are accessable via each other (internal SP network rechability)

Hi Vivek,

Thank you for the reply.

Next question is, how then the P routers forward the packets (example below, CE left side ping CE right side) if they don't know those networks and according to the routing process the destination IP never changes along the path only layer-2 address is swapped.

As far as I know MPLS is adding label only to prefixes exist in the routing table and also the LSP is build based on the RT.

 

CE1 ping 10.0.0.1 --> PE --> P1 -> P2 -> P3 -> P4 --> PE --> CE1 IP: 10.0.0.1

 

Please correct me if I'm wrong

Hi,

MPLS is working in layer2.5 (in between Data link and Network layer of OSI model). PE routers map the prefixes received from CE into labels and at the P routers it just have to swap the labels, no lookup based on IP/Layer3 information. VRF will also add an additional label and used in case of multiple CEs and inter-customer routings.

That is the basics of MPLS and its hard to explain in a single post. You need google to understand in more details. Below are few example that may help you.

 

http://www.cisco.com/web/CA/events/pdfs/Intro_to_mpls.pdf

 

http://www.cisco.com/c/en/us/td/docs/net_mgmt/vpn_solutions_center/2-0/mpls/provisioning/guide/PGmpls1.html

 

Hi Vivek, thanks for the reply.

After spending a few hours in GNS3/search I found the answer.

When a CE advertise a route in the PE router, PE router creates a VPN-ipv4 address with a related VRF/route distinguisher/route target/VPN label (all these are pre-configured manually). PE routers in turn inject those routes in the IBGP (to the others PE ONLY).

The part of the puzzle I missed was the VPN label (CE prefixes) which is injected as the bottom label (stack set to 1) in the MPLS header and is used only by the PEs and never checked by the Ps. All Ps in the middle are using only the top label for the data-plane/transit-packet which is usually associated with a loopback on the other PEs.

The FEC/LSP is created by the PE routers in control-plane and the forwarding between the P routers is made in the data plane (LFIB).

In short, P routers have no clue about the CE prefixes.

Cheers

 

Posts in this discussion have been modified due to possible misconduct. Please refer to the CSC terms of use for more details. 

Getting Started

Find answers to your questions by entering keywords or phrases in the Search bar above. New here? Use these resources to familiarize yourself with the community: