MPLS

[keven.jones]

Hi Group,

I got a basic question.

A, B, C, D are four different locations connected via MPLS Cloud provided by telecom-provider.

All location got OSPF routing protocol and to inter-connect locations telecom-provider provides BGP config and neigh IP

Telecom config

router bgp 3501

neighbor  10.10.3.209 remote as 3500

no auto-summary

To access resources from anywhere to anywhere do I need mutual redistribution between OSPF and BGP on all location.

Please Help!

cheers

Keven

[Richard Burts]

Keven

Without knowing more about your network environment it is difficult to give really good advice on your question.for example, do these sites have multiple routers and multiple subnets at each site or are all subnets present on the router that talks to the ISP? Is the ISP advertising a default route via BGP to each site? Do the sites have Internet access independent of the MPLS connectivity?

But without knowing those details I believe that it sometimes would be possible to run OSPF as the routing protocol within the site and BGP as the routing protocol to the ISP (and therefore between sites) and not have mutual redistribution.

If the ISP BGP is advertising a default route to the site, then it should be possible to configure the site OSPF router that connects to the ISP witth default-information originate and not require redistribution from BGP into OSPF.

If the subnet routes for the site are present in the routing table of the site router that connects to the ISP (and especially if there has been summarization of the subnet routes or there is a null 0 route for the block of subnets at the site) then it should be possible to use a network command in BGP to advertise the routes from the site without requiring redistribution from OSPF into BGP.

HTH

Rick

[keven.jones]

Hi Richard

Some background on the network. NO Internet is involved on any location. Each location got multiple subnets.

On the site router of each location I can see other location subnetss using "show ip route bgp". Each location have site router terminating MPLS and a Layer3 Switch acting as BackBone. Point-to-Point link will be terminated on L3 Switch

Location A and Location B are across the street and will soon be linked through Fiber, hence only these two location will have Point-to-point link and MPLS Link. Please suggest where I need mutual redistribution and how do I control flow of traffic and updates between location A and location B as they got more than one connection.

cheers

Keven

[Jon Marshall]

Keven

The way we did a similiar setup was to -

1) redistribute BGP into OSPF (EIGRP in our case but little difference)

2) advertise each sites networks using the "network" command under the BGP config. If you can summarise each site then even better and then you can use the aggregate-address command under the BGP config.

Alternatively you can simply add a route for the summary route on your BGP speaking router to Null0 eg.

ip route x.x.x.x Null0

the x.x.x.x would be a summary route for all that sites networks. You need to have a matching route in the IGP routing table for BGP to advertise out a "network" statement and this is how you would do it. Nothing should actually get routed to Null0 because the same router should have the more specific routes for each subnet in it's routing table.

Jon

[keven.jones]

Hi jon

Unfortunately summarization not possible on locations A,B,C,D.

• Location A and Location B got two routes to same desitnation which one it would prefer ( in case of location A to location B )
• Location A and Location B can get updates from Point-to-Point and MPLS , which one it would prefer?
• On Location A and Location B Layer3 Switch (core) and site MPLS router - Could they get into loop for exchanging updates

Can you demonstrate a sample config on your input.

cheers

keven

[Mahesh Gohil]

Hi Keven,

What I understood is

> All your locations have MPLS link to rovider
> Two locations are connected by p2p link in addition to MPLS link
> MPLS link running bgp with provider
> p2p link is having ospf as protocol

for A and B you have two choice to reach other. via MPLS link (BGP: Administrative distance:20) and p2p link (OSPF: Administrative distance:110) so by default it will prefer mpls link due to low admin distance.

Now config:

For siteA

router bgp
neighbor remote-as
redistribute ospf match internal external 1 external 2

router ospf
redistribute bgp
network
network

One more thing as I discussed above If you want your traffic t move via. p2p link and not mpls link you can change the distance of bgp to higher one so that p2p link will be prefered

router bgp
distance bgp 120 200 200

As you can see we increase admin distance of bgp to 120 (higher than ospf:110) so p2p link will be prefered.

you can repeat the same at siteB

Hope this helps

Regards
Mahesh

[keven.jones]

Hi

Thank you for supporting.  Setup goes as

Loc A (3560)  --------------------------p-2-p -------------------------Loc B (3560 )

Loc A (3560) -----------2800_MPLS_RTR_A------------(MPLS provider)----------1841_MPLS_RTR_B----------loc B (3560 )

There will be two point of mutual redistribution, i.e on location A and location B

Also Location A can get updates from point-to-point link and MPLS link, will this not cause looping

cheers

Keven 

[Mahesh Gohil]

Hello Keven,

It will not create any loop because though you have two exit point ,at a time only one will be prefered.

Regards

Mahesh

[keven.jones]

Hi Mahesh

I got your point.

Traffic originating from Loc B to Loc A prefers MPLS rather than P-2-P link once mutual redistribution is done.

Where do I need to modify the AD on Loc B ( Backbone router or MPLS router )

cheers

Keven

[Mahesh Gohil]

Hello Kevan,

it is on redistribution point, your MPLS router

Regards

Mahesh

[keven.jones]

Hi Mahesh,

Location A requires specific subnets to be advertised to its neighbors and accept specific subnets from its neighbor.

To fullfill the above need, do we need to apply the filter within BGP process or with redistribute command.

option I

router bgp 5

neighbor 100.1.2.2 prefix-list out-peer out
neighbor 100.1.2.2 prefix-list in-peer in

Option II

router bgp 5
redistribute ospf 9 match internal external 1 external 2 route-map control-ospf

Cheers

Keven

[Mahesh Gohil · Accepted solution]

Hello keven,

Option-1 will restrict specific subnet to advt. and specific subnet blocked at your MPLS router.

Option-2 will restrict specific network blocked from advt. but will not block any network coming to your MPLS router. To do both

              you need to apply route-map while distributing bgp routes to ospf

              router ospf 9

              restribute bgp 5 route-map BGP-TO-OSPF

BGP is more handy in controlling in/out prefixes via. different attributes

Hope this helps

Regards

Mahesh