Cisco Support Community
Showing results for 
Search instead for 
Did you mean: 
New Member

OSPF Routing issues with WAN triangle

We are in the process of completing a trangle between our three main sites. Today we are a hub/spoke design using OSPF with Area 0 being at our main site and Area 1 and 2 at the other two. When we add the links between the other areas we need to decide on how best to deal with the area0 issues. I've read here that some move area 0 to the WAN, other say not to. I've looked at redistributing OSPF into EIGRP or BGP for the WAN connections. Will this allow us to use the links between the two non-area0 sites? If not am I going to need to put OSPF on the WAN?

The WAN will consist of two high speed circuits between each of the three sites.



New Member

Re: OSPF Routing issues with WAN triangle

Hi John,

probably the least complicated method is to advertise the WAN links in area 0. This will effectively make each router in the "triangle" an ABR. Not a big concern but ensure that you have enough processing power and memory in these routers. Being main sites, I assume they would be suitably sized.

Any direct link between Area 1 and Area 2 should also be in Area 0 - if not, you will be back to a hub/spoke scenario based on the preference of Area 0 links over any other area.



Hall of Fame Super Silver

Re: OSPF Routing issues with WAN triangle

Hello John,

if your WAN cloud is made of fast and reliable high speed links just put them in OSPF area 0 because there OSPF will work well on them.

So I agree wih Steve

Hope to help


Super Bronze

Re: OSPF Routing issues with WAN triangle

As the others have suggested, place the three main sites in area 0. However, further suggest making the existing main site area 0 into a new ospf area. I.e. the backbone would comprise the mesh WAN connections of the three main sites, each of those site would contain its own local area.

New Member

Re: OSPF Routing issues with WAN triangle


You should also consider why your remote sites were put into different areas in the first place. Most likely it was to allow you to summarise at the borders: for example if you have a lot of routing changes in one area they can be masked from being sent over the WAN by using stub areas. It may be that your network is just of a large enough size to warrant more areas - the point is that once you change all areas to area 0, you are likely to have more ospf transmissions across your wide area.

If you're connecting your area 1 - area 2 together you also need to consider the situation if either area 1 or area 2 lose their direct connection back to Area 0. In this case you would need a virtual link back to area 0 to ensure that all areas are connected back to area 0.

You mention the use of BGP - another option is to redistribute your OSPF table at each border into BGP and use BGP to transmit your routes to the remote site. In this case, all the other sites could be set to area 0 as they will be isolated from each other - although they would run perfectly happily as a single non area 0 area.



CreatePlease to create content