cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
2254
Views
0
Helpful
5
Replies

Condictional Route Advertisement - RIP / OSPF / EIGRP

guruprasadr
Level 7
Level 7

Hello Experts,

Scenario: Consider T3 PoPs connected to nearest T2/T1 PoPs via the DS3 or Metro Ethernet Link.

If the reach ability to Network Core is Lost for the T3 PoP, my Customers whose last mile (landing on T3 PoP Router) having Roting Protocol as RIP / BGP / OSPF / EIGRP, the "default-originate" configured at T3 Router is still continuing to originate the default 0.0.0.0 for CE.

Customer's Secondary Link will not Kick-off unless & until the PRY LAST MILE is going DOWN. During this situation, the Customers are impacted & we need to wait until the core reach ability link is restored (or) manually disturb the PRY Customer Link.

In BGP, we are manipulating the case with "BGP Conditional advertisement" using command: "neighbor 10.0.7.6 default-originate route-map"

Question is:

In RIP / OSPF / EIGRP as the LAST MILE routing protocol for the Customers. How to manipulate this case/situation ?

Request your Help/Support in Advance.

Best Regards,

Guru Prasad R

5 Replies 5

Giuseppe Larosa
Hall of Fame
Hall of Fame

Hello Guru,

if using two different protocols with OSPF you can create a default conditioned on receiving one default route or other prefix from the DS3 link

ip prefix-list only-default permit 0.0.0.0/0

access-list 11 permit host bgp.nexthop

route-map check-ebgp-default permit 10

match ip address prefix-list only-default

match ip next-hop 11

set metric 500

router ospf 10

default-information originate route-map check-ebgp-default type 1

if the eBGP session goes down or the eBGP session is lost OSPF stops to advertise a default route downstream

We used this solution with two internet gateways.

For EIGRP and RIP you could think of the ip default network command (that requires that network to be in a different major network).

This network must be in the routing table otherwise the default route is not propagated.

you can achieve this with a proper readdressing of T3 POP to DS3/ metro ethernet

Hope to help

Giuseppe

HI Giuseppe,

Thanks for your immediate reply.

Sorry, i did not catch the exact point in your configuration.

"access-list 11 permit host bgp.nexthop"

"match ip next-hop 11 "

"set metric 500"

What is the actions/effect of the above commands.

Again, my IGP is OSPF, ie., OSPF is established between the T3 - T2/T1 Routers as uplink.

As per your configuration, the tracking is for the BGP.nexthop ip address. How do i track (or) do a condictional default originate in-case my OSPF as the IGP Protocol.

Thanks in Advance.

Best Regards,

Guru Prasad R

Hello Guru,

If I understand correctly your topology:

you use a single OSPF routing domain:

you have some regional site/POPs that act as concentrators for other T1/T2 link connected branch sites.

The T3 concentrator site is connected to the core via DS3 or metro ethernet.

If I have understood your first post your T3 site router is injecting a default route towards the T1/T2 branch sites.

Branch sites have an alternate path to reach the core but this isn't triggered until the direct link between T3 site and the branch site goes down because the default route is generated by T3 router unconditionally.

I guess the T3 router does this because it is an ABR between backbone area and another area that can be some form of stub area where the all the links to the branch and the branch routers are placed.

If the T3 link to the core goes down the T3 router keeps advertising that default route in the regional OSPF area.

This doesn't allow the branches to trigger the desired backup action.

A design suggestion wuould be that of providing an alternate path to the T3 ABR router.

Another possible solution can be the following:

have the ABR T3 router injects a summary route into the regional area as a consequence of LSAs received on the area 0 link.

You can filter all the other O IA routes using the area filter-list command.

The regional area has to be made stub not totally stub in order to be able to accept O IA in addition to the default route.

On the branch routers you can use the dialer watch feature with any type of backup using DDR: ISDN but also PPPoE as well.

The dialer watch feature allows to monitor specific routes in the routing table: when the summary O IA routes vanish in the routing table it is sign that T3 router core link has failed.

This event is used to trigger the backup action

for dialer-watch see

http://www.cisco.com/en/US/docs/ios/12_2/dial/configuration/guide/dafbakdw.html#wp1000976

Edit:

I tried to see if an NSSA area where you explicitly configure the ABR to inject a route with

area area-id nssa [no-redistribution] [default-information-originate [metric] [metric-type]] [no-summary]

but there isn't an option to reference a route-map so my understanding is that the default route if set is sent always.

see

http://www.cisco.com/en/US/docs/ios/iproute/command/reference/irp_osp1.html#wp1011294

Edit2:

if ospf is only used in the last mile you can use an approach as described in my first post just you need to adapt the criteria used inside the route-map

Hope to help

Giuseppe

HI Giuseppe,

Sorry about the late reply.

As per your advice,

"For EIGRP and RIP you could think of the ip default network command (that requires that network to be in a different major network).

This network must be in the routing table otherwise the default route is not propagated"

Clarifications:

===============

We are implementing the EIGRP & RIP as the last-mile routing protocol for the MPLS-VPN Customers.

Normally my default network on the Service Provider Edge Routers will be on OSPF.

How do i manipulate now, if the reach ability to the Core is lost. In this case, my Default route should be with-drawn and the Secondary Link should trigger.

Kindly advice.

Best Regards,

Guru Prasad R

Hello Guru,

no problems

If I understand correctly:

the scenario is that of a L3 MPLS VPN service.

the PE-CE protocol is RIP or EIGRP (last mile routing protocol)

the CE receives a default route in the IGP of choice (RIP or EIGRP).

This default route should be ideally conditioned to the good connectivity to core of PE router(s).

option1:

within the VRF context the PE routers receive a default route from some other PE (like in internet access for MPLS VPN).

the PE within the context of the VRF redistributes BGP AF vrf vrf-name into EIGRP /RIP.

(to be verified: this happens for the default route too ?)

the PE can also receives and use the details of routes of all other VRF sites taking part in the same VPN.

In this way the received default route is never used by PE for routing VPN traffic but just for having a way to send a conditional default route to CE router.

This requires that redistribution of BGP into EIGRP uses a route-map with an access-list/prefix-list to match only default-route.

The CE router can use a floating static route approach:

the primary default route is that received from RIP/EIGRP.

the secondary default route points to the correct dialer interface for DDR over ISDN or DDR using PPPoE.

When connection to the backbone is lost:

the PE router stops to consider valid the iBGP route in VRF for net 0.0.0./0

The PE router removes the RIP/EIGRP advertisement of net 0.0.0.0/0

The CE triggers the usage of DDR.

This should be tested in a lab. The critical point is to see if it is possible to pass the default route from BGP to RIP/EIGRP router.

Alternative:

RIP and EIGRP allows the creation of summary-address out a specific interface:

the summary-address is sent out if a component subnet exists in the RIP/EIGRP DB.

in this case the PE redistributes all the routes from BGP to RIP/EIGRP.

out PE-CE interface a manual summary is sent.

the manual summary is removed when no component subnet is present in the DB (the DB is emptied when connectivity to the backbone is lost)

on the CE a similar floating static route for the same summary route is configured using the dialer interface.

Hope to help

Giuseppe

Getting Started

Find answers to your questions by entering keywords or phrases in the Search bar above. New here? Use these resources to familiarize yourself with the community:

Innovations in Cisco Full Stack Observability - A new webinar from Cisco