Question MPLS CE Configuration

Unanswered Question
Aug 20th, 2007
User Badges:

We have set up a new MPLS link to one of our remote offices and I have some configuration questions. We have 3 sites. Site A is the HQ. Site B and C are remote offices. Site A and B are connected through a DS3 Point2Point line. Site A and Site C are connected through the new MPLS line. (See attached drawing). Internally we are using EIGRP to handle routing between data and voice networks. Now that we have the MPLS line I am not sure on how to go about setting up EIGRP. Site C will need to be able to reach both Site A and B on both the voice and data networks. I did some searches and found that typically BGP is redistributed into EIGRP. I came up with the following config:


****************

Router A


router bgp 65xxx

no synchronization

bgp log-neighbor-changes

network 192.168.100.0 0.0.3.25

neighbor 12.x.x.x remote-as xxxx

no auto-summary


router eigrp 200

network 192.168.100.0 0.0.3.255

network 172.16.0.0 0.0.0.3

redistribute bgp 65xxx metric 1500 2000 255 1 1500 route-map BGP-to-EIGRP


route-map BGP-to-EIGRP permit 10

match ip address 10


access-list 10 permit 192.168.4.0 0.0.0.255

access-list 10 permit 10.10.4.0 0.0.0.255



***********

Router B


router eigrp 200

network 192.168.6.0 0.0.0.255

network 172.16.0.0 0.0.0.3

no auto-summary



**********************

Router C


router bgp 65xxx

no synchronization

bgp log-neighbor-changes

network 192.168.4.0 0.0.0.255

neighbor 12.x.x.x remote-as xxxx

no auto-summary


router eigrp 200

network 192.168.4.0 0.0.0.255

redistribute bgp 65xxx metric 1500 2000 255 1 1500 route-map BGP-to-EIGRP


route-map BGP-to-EIGRP permit 10

match ip address 10


access-list 10 permit 192.168.100.0 0.0.3.255

access-list 10 permit 10.10.100.0 0.0.3.255

access-list 10 permit 192.168.6.0 0.0.0.255

access-list 10 permit 10.10.6.0 0.0.0.255



The Catalyst switches are also running EIGRP so currently Site and B are able to communicate on both the data and voice networks.



What else do I need to be able to route between all three sites on both data and voice networks?



Attachment: 
  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 0 (0 ratings)
Loading.
mohammedmahmoud Mon, 08/20/2007 - 23:24
User Badges:
  • Green, 3000 points or more

Hi,


Why using BGP on your side, you can simply ask your provider to use EIGRP as the PE-CE routing protocol between site A and C, and everything will go fine (what the provider does is that on the first PE he redistributes your EIGRP routes into his internal BGP until reaching the other PE and then the routes are redistributed back into EIGRP and are sent to your CE and vice versa).


HTH,

Mohammed Mahmoud.

Jon Marshall Mon, 08/20/2007 - 23:58
User Badges:
  • Super Blue, 32500 points or more
  • Hall of Fame,

    Founding Member

  • Cisco Designated VIP,

    2017 LAN, WAN

Hi


Okay quite a few things. Assuming your provider does not support EIGRP across their MPLS network


1) Your netmasks under BGP config need modifying as BGP does not use inverse masks so


network 192.168.100.0 0.0.3.255


should be


network 192.168.100.0 mask 255.255.252.0


etc.. for all your network statements under BGP.


2) It's not clear from your schematic how A & C are peering. They will not peer with each other but with your providers PE routers.


3) You will need to have network statements on Router A under the BGP config for all the data and voice vlans for both site A and site B.


4) Do you know if the AS number is the same at A and C which is quite common. if it is you will need to add the following line under bgp config on router A & C


neighbor x.x.x.x allowas-in 1


5) The network statement on Router A ie.


192.168.100.0 255.255.252.0 + the voice one that is not there ie.

10.10.100.0 255.255.252.0


Are these summary routes are are they actual subnets ?.


if they are summary routes for BGP to advertise them out you need either


i) A route in the routing table that exactly matches your network statement. So if this is a summary route you probably won't have a match in the routing table. The way to get a route into the routing table is to create a static route for the summary route pointing to Null0.


ii) use the aggregate-address command under the BGP config.


ii) is preferable in my opinion.


HTH


Jon


Edit - oops sorry Mohammed i guess you must have posted that while i was writing mine.



mohammedmahmoud Tue, 08/21/2007 - 01:26
User Badges:
  • Green, 3000 points or more

Hi Jon,


No problem at all :)


But don't you think that using EIGRP as PE-CE routing protocol is the better solution for him.


BR,

Mohammed Mahmoud.

royalblues Tue, 08/21/2007 - 01:31
User Badges:
  • Green, 3000 points or more

I think it would depend on which protocol he is most comfortable with.


I would prefer BGP as this gives us a lot flexibility in terms of enforcing policies :-)


Narayan

Jon Marshall Tue, 08/21/2007 - 01:33
User Badges:
  • Super Blue, 32500 points or more
  • Hall of Fame,

    Founding Member

  • Cisco Designated VIP,

    2017 LAN, WAN

Hi Mohammed


It could well be as it really is a very small network setup but i think the problem may be that the provider might not support EIGRP. Not sure about worldwide but here in UK we were really only given the options of static routing or BGP.


We chose BGP for precisely the reason Narayan gives.


Jon

mohammedmahmoud Tue, 08/21/2007 - 01:47
User Badges:
  • Green, 3000 points or more

Hi Jon and Narayan,


Jon, as you said it is a simple network setup and thats why i think EIGRP would be better than BGP, where i am working (SP in Egypt) we offer RIP, EIGRP, static and BGP, as using RIP or EIGRP would be more handy for most of the Customers' administrators, while BGP has extra admin burden (Narayan, not all people are looking for enforcing policies :) ).


BR,

Mohammed Mahmoud.

Mohamed Sobair Wed, 08/22/2007 - 04:12
User Badges:
  • Gold, 750 points or more

Hi Mohamed,


I am working in SP (Saudi Arabia) and we really offer (bgp + static) , and most of the cases we are running BGP with the customer since our MPLS core (P) and PE's are Juniper routers and do normal redistribution between BGPv4 customer addresse to our MPLS VPNv4 cloud and imported back.


Regards,

Mohamed Sobair


Actions

This Discussion