BGP default route Originate

[Vinayaka Raman]

 Topology

ISP------- ebgp-----------------HUB site R1--------ebgp-----------MPLS-------ebgp---------remote sites       

                                                  | ibgp                                      |

                                           HUB site R2      ------ebgp---------- 

ISP is injecting a default route to R1 and i could not advertise this to remote site. The reason is ISP AS number and MPLS AS number are are same.

I used conditional route-map i.e, neigh default originate with route-map command to inject a default route from R1 to remote sites via MPLS cloud.

It works fine. But the problem is i could not set community tag to this default route using a outbound route-map at R1. The reason being it is route orginated by default originate command and route map becomes ineffective. I would like to use community attribute because in future we will be injecting one more default route from a different hub site into the same MPLS cloud. Then some sites has to prefer HUB1 and some other to HUB2 for internet traffic.

Any suggestions?

if R1 is learning is default route from ISP and it is in the local routing table, just network 0.0.0.0 should inject a default route to MPLS cloud? same as case in R2..but this isn't working for me.

[Amit Goyal]

Hi Vinayaka,

I proposed below solution which i have implemented many times.

ISP------- ebgp-----------------HUB site R1--------ebgp-----------MPLS-------ebgp---------remote sites       

                                                  | ibgp                                      |

                                           HUB site R2      ------ebgp---------- 

>On HUB Site R1 configure interface towards ISP and MPLS provider in vrf.

>Activate the ebGP session in same vrf address-family.

Configure egress as-override feature with below command on HUB R1.

router bgp <AS>

address-family ipv4 vrf <vrf name>

neighbor <mpls neighbor ip> as-override

With the above config in place AS path info coming from ISP will be override with HUB AS while prefix being sent out to MPLS.

HTH

-Amit

[Vinayaka Raman]

Can we enable as-overide without vrf ?

[Amit Goyal]

No you can not since it is a feature used by PE devices at provider end. Specially used for CE which are part of vrf.

HTH

-Amit

[Vinayaka Raman]

if i had to enable as-overide i have to do it on hub site R1 towards the mpls peer and also the internet peer. this is because  need to learn default route from internet peer and advertise to remote site and also learn public prefixes from remote site and advertise it to internet peer. 

right now i am using static to bgp redistribution at R1 to advertise to internet and default originate to inject default route to remote sites.

is this VRF going to creating another routing table instance? i am worried how many complication it will introduce..

[Amit Goyal]

>if the routes are flowing in both direction then yes "as-override" is required for both neighbors.

>Yes there us going to be another routing table instance.

>If possible check with ISP and MPLS provider to configure "allowas-in"

HTH

-Amit

[Vinayaka Raman]

Hi Amit,

one question,

AS 9874 internet PE --------------AS 65000 CE-----------PE AS 9874 mpls cloud

if i request my provider to enable as-override on all the MPLS PE, will it override the immediate AS (65000) only or all the AS behind (internet AS 9874 also) ?

[Amit Goyal]

Vinayaka,

As-override is a egress feature. With this feature enabled, a BGP speaking device will remove all AS-Path from the string and add its own AS only while advertising the prefix to eBGP peer.

In your case even if MPLS provider enable this feature but it will not help since the default route will be denied during the advertisement from CE to MPLS PE.

You may need to ask your ISP to configure allowas-in.

HTH

-Amit

[Amit Goyal]

If your MPLS provider is ok to configure "allowas-in" for HUB R1 then it can be other solution of this design.

HTH

Amit