LDP BGP routes distribution

[rac.vishaldhingra]

Hi,

In my routing table I have

1. OSPF routes

2. Local Routes

3. BGP Routes(Suppose total 10 BGP : 2 of them I have re-distributed and 8 of then are not re-distributed)

I have a MPLS LDP CISCO 7200 router.

By default LDP will not distribute BGP routes.

Problem statement :

I want to distribute label for 2 BGP routes which are re-distributed.

How can I do so ?

Should I go for Prefix Lists for MPLS LDP Local Label Allocation Filtering ?

But seems to be static mechanism.

Please tell.

Regards,

Vishal

[Giuseppe Larosa]

Hello Vishai,

in order to achieve this that is contrary to common use of MPLS the routes have to be redistributed into BGP and LDP local label allocation fiiltering has to be updated on all device not only on the PE that originated the OSPF external route.

To redistribute from iBGP into OSPF you need a command under router bgp.

router bgp xxxx

bgp redistribute-internal

!

see

http://www.cisco.com/c/en/us/td/docs/ios-xml/ios/iproute_bgp/command/irg-cr-book/bgp-a1.html#wp4270480859

Use a route-map to decide which routes are redistributed from iBGP into OSPF and then update the  LDP Local Label Allocation Filter

However, it is not clear why you want to have the two BGP routes have their own LDP LSPs as they are already going over the LSP with destination the remote PE loopback that must be the same as the MP i BGP next-hop in global routing table (GRT)

The only advantage is to have separate byte counters for the new dedicated LSPs in the show mpls forwarding table.

But you can use netflow version 9 to achieve the same results in a more scalable way.

Hope to help

Giuseppe

[Adam Vitkovsky]

What are you trying to achieve please?

If you say “ in my routing table I have OSPF, local and BGP routes” do you mean global routing table or VRF routing table.

If these routes are in global routing table:

Cisco implementation of LDP will by default allocate and distribute labels for all routes in global routing table.

If these routes are in a VRF routing table:

Then they don’t need to have LDP labels –the next-hop of the route (in global routing table) and it’s label is used instead.

If you’re trying to run LDP in a VRF, then that’s a whole another story.

adam

[Giuseppe Larosa]

Hello Adam,

>> Cisco implementation of LDP will by default allocate and distribute labels for all routes in global routing table.

BGP routes will use BGP next hop recursion so they re-use LDP LSPs with destination the BGP next hop of the route.

What you say is true for IGP routes only in my working experience.

However, thanks for your contribute in this and other threads of MPLS forum it is a pleasure to see other members helping in answering questions.

Best Regards

Giuseppe

[Adam Vitkovsky]

Hello Giuseppe,

>> Cisco implementation of LDP will by default allocate and distribute labels for all routes in global routing table.

>BGP routes will use BGP next hop recursion so they re-use LDP LSPs with destination the BGP next hop of the route.

>What you say is true for IGP routes only in my working experience.

You’re right, my mistake, yes indeed the recursion not only happens for VRF BGP prefixes but also for GRT BGP prefixes.

Regarding the allocation of transport labels for BGP routes, now that I think about it OP can use RFC 3107 to allocate and distribute labels for BGP prefixes.

adam

[hunterlau]

Hi Giuseppe,

Why BGP routes select MPLS LSP instead of plain IP? I cannot find the behavior of Cisco, but I found the behavior of juniper.

"BGP selects next-hop entries in the inet.3 routing table because their preferences are always lower than OSPF and IS-IS next-hop preferences. "

https://www.juniper.net/documentation/en_US/junos/topics/concept/mpls-routing-tables.html

Thanks.

[Harold Ritter]

Hi,

There is no such concept in IOS or IOS-XR. By default, the best path is selected, whether MPLS is enabled on it or not. In a service provider network, all the links will either be all MPLS or all native IP and in the case of an MPLS network, all links will be MPLS and the case where a link does not have MPLS enabled is more likely a misconfiguration than the intended configuration.

To ensure that all links are MPLS enabled, you can configure a feature called LDP IGP Synchronization, which will cause the IGP to assign a very high metric to a link where LDP is not configured or not functioning properly.

You can also configure "mpls ldp autoconfig" to configure LDP automatically to all links included in the core IGP.

Regards,