cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
1218
Views
0
Helpful
2
Replies

OSPF Failover design / Tunnels

roger perkin
Level 2
Level 2

I am trying to design a solution to enable the failover of our data centre to use a backup MPLS link.

Please see attached jpg for a basic diagram.

The primary connection between the Data Center and the DR site is a dedicated site to site link.

6509 at each site peering ospf all in Area 0

we also peer ospf with the MPLS backup network routers

The 10.50.0.0 address space is specified in a network statement in BGP on the MPLS router at data centre site

The 10.51.0.0 address space is sepcified in a network statement in BGP on the MPLS router at the DR site

This is then redistributed into our OSPF with a high metric.

The site to site link is 1GB and is the preffered path for traffic to the DR site.

During testing we take down the site to site link and routing to the 10.51 network works over the MPLS fine

We are now adding extra networks at the data centre network and I have a few questions which I need to understand more.

1. When the site to site link goes down this leaves OSPF area 1 hanging ? Is this a problem

2. Should I be looking to run a tunnel between the two MPLS routers to ensure complete connectivity for our OSPF domain.?

Does anyone have a designs similiar to this or can point me to any best practice documentation.

Thanks

Roger

2 Replies 2

Giuseppe Larosa
Hall of Fame
Hall of Fame

Hello Roger,

if the backup link is based on a L3 VPN and you are using BGP as PE-CE protocol, it shouldn't be a problem to leave OSPF area 1 alone in failure scenario.

>> The 10.50.0.0 address space is specified in a network statement in BGP on the MPLS router at data centre site

The 10.51.0.0 address space is sepcified in a network statement in BGP on the MPLS router at the DR site

1)  only impact is on ABR device in DR site you will not be able to reach IP addresses in area 0 as it will be in a partiotioned area 0.

for example if ABR loop0 is in area 0.0.0.0 during failure you will not be able to reach it unless it is distributed into BGP

2) no need for a GRE tunnel just to join the partitioned area 0.0.0.0 if you are using BGP on the backup link

Hope to help

Giuseppe

Thanks Giuseppe,

I have labbed this up in GNS3 and it does not appear to cause any issues.

I was concerned about breaking Area 0 in half and as to the behaviour of that.

But from what you are saying, it is not a big concern.

Roger