cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
1121
Views
5
Helpful
4
Replies

Native IP traffic out of LSPs with 'autoroute-announce' configured

Dear all,

I'm building an MPLS lab with 7600s and ASR1000s routers. Up to now, I already have some VPLS and pseudowires configured, apart from native IP traffic. I use OSPF for internal addresses (loopbacks and internal links) and iBGP for 'external' addresses, the ones of the emulated customers. LSPs don't have 'traffic engineering autoroute-announce' configured, so IP traffic is natively routed as LSPs are not considered valid interfaces by OSPF. Now, I would like to configure L3VPNs and I need to configure "autoroute-announce" (or static IP routes for the routers' loopback interfaces). If I'm not wrong, this will imply that IP traffic will start to traverse the LSPs. Is there any way to keep native IP traffic out of the LSPs once 'autoroute-announce' is enabled?

Thanks in advance

Octavio

4 Replies 4

Harold Ritter
Cisco Employee
Cisco Employee

Hi Octavio,

I have a few questions for you. VPLS does require that you have LSP setup between PEs. You seem to indicate that you have TE tunnels but that you don't use them. Do you use LDP to setup LSPs for VPLS?

Regards

Harold Ritter
Sr Technical Leader
CCIE 4168 (R&S, SP)
harold@cisco.com
México móvil: +52 1 55 8312 4915
Cisco México
Paseo de la Reforma 222
Piso 19
Cuauhtémoc, Juárez
Ciudad de México, 06600
México

Hi Harold,

Thank you for your help. I didn't explain everything properly. I try to elaborate a bit more on the architecture:

* LSPs for pseudowires and VPLS services are set up by RSVP-TE. There is a full-mesh.

* pseudowires and VPLS services are signalled by T-LDP sessions and they work properly

* TE information is sent by OSPF (opaque LSAs)

* OSPF is used for propagating infrastructure's IP addressing (loopbacks and internal interfaces between routers)

* BGP is for propagating IP addresses assigned to the routers emulating IP transit customers

IGP doesn't include LSPs in the shortest path calculation as LSPs are configured this way:

interface Tunnel3001

ip unnumbered Loopback1

tunnel mode mpls traffic-eng

tunnel destination X.Y.C.Z

tunnel mpls traffic-eng path-option 10 dynamic attributes Tunnel3001

!

So, that's way IP transit traffic is natively routed.

To enable L3VPNs, if I'm not wrong, I need reachability of the routers' loopback interfaces (source of the MP-BGP sessions) through the LSPs and I get it by static routes or configuring tunnels this way:

interface Tunnel3001

ip unnumbered Loopback1

tunnel mode mpls traffic-eng

tunnel destination X.Y.C.Z

tunnel mpls traffic-eng autoroute announce

tunnel mpls traffic-eng path-option 10 dynamic attributes Tunnel3001

!

As these loopbacks are also the source of the iBGP and now the IGP will prefer to reach them through the LSPs, I assume the IP traffic will be sent through the LSPs. I wanted to know if there is a config that at the same time allow me to configure L3VPNs and have the native IP traffic out of the LSPs.

Hope this can help you to better understand the scenario.

Thanks a lot for your help

Kind regards

Octavio

Hi Octavio,

You could use different loopback interface addresses for your ipv4 and vpnv4 iBGP sessions, map static routes for the vpnv4 loopbacks to the TE tunnels and leave the loopback addresses used for ipv4 transit be routed unlabelled via the IGP.

Why is it that you do not want the IPv4 transit traffic to be label switched? The easier solution woul be to continue to use RSVP TE for VPLS if this is a requirement and to use LDP signalled LSP for everything else (l3vpn and ipv4 transit).

Regards

Harold Ritter
Sr Technical Leader
CCIE 4168 (R&S, SP)
harold@cisco.com
México móvil: +52 1 55 8312 4915
Cisco México
Paseo de la Reforma 222
Piso 19
Cuauhtémoc, Juárez
Ciudad de México, 06600
México

Thanks a lot, Harold. It's clear your solution and, for sure, it will work.

To be honest, I prefer to label switch IPv4 transit traffic. However, in our production network we have routers from another vendor that, by default, route natively IPv4 traffic, unless you explicitily configure 'IGP shortcuts' (equivalent to 'autoroute announce'). I'm sure I will be asked about the possibility of emulating this behaviour in the Cisco part of the network and I wanted to check in the lab that it was possible in fact.

Harold, I really appreciate your help. Thank you very much.

Octavio