Same network in two different OSPF areas, possible?

[Mohammed Islam]

Hi,

I'm trying to find out whether there is any implications to having a SAME network/subnet in TWO different OSPF areas.  If so, what could be the issues and are there any workarounds?  This is for a DR scenario.

Scenario:-

HEADEND

OSPF 1

network - 10.100.0.0 area 0

  |

VPN

  |

FAR END

OSPF 1

network 11.0.0.0 area 10

network 10.100.0.0 10 <<-------

This is so that if a particular device goes down in area 0 (headend) that device can be fired up in area 10 (far end).

Appreciate your assistance. Thanks.

[cadet alain]

Hi,

you won't have an adjacency as the neighbours are not in the same area.

I supposed the 10 network is the VPN network between the 2.

Regards.

Alain.

Don't forget to rate helpful posts.

[Mohammed Islam]

Thanks for the response.  To be a a bit more clearer:-

HEADEND

Switch/Router

OSPF 1

network - 10.100.0.0 area 0

network - 172.1.1.0 area 10

  |

172.1.1.0/30

FW area 10

172.1.1.4/30

  |

VPN

  |

FAR END

172.1.1.8/30

FW - area 10

172.1.1.12/30

|

Switch

OSPF 1

network 172.1.1.0 area 10

network 11.0.0.0 area 10

network 10.100.0.0 area 10 <<-------

OSPF adjacency is fine as it forms on the 172.x.x.x network.  The question is really around having the 10.100.0.0 network in two difference areas.  What would the issues be if any?

[cadet alain]

hi ,

if network 10 is on one router in area0 where it is physically connected then it will appear on the other end as coming from area 0 and the second command won't have any effect if this network is not physically on the second router

Do I understand correctly your problem ?

Regards.

Alain.

Don't forget to rate helpful posts.

[Mohammed Islam]

The network will be physically on the second/remote router/switch with devices connected to it.  This is idea:-

Main 10.100.0.0 network - devices plugged in.  If these devices go offline etc, ,there is a clone of these remotely in the far end.  The clones NEED to have the SAME ip as their original counterparts and need to talk/route to to the headend to a few servers.

[smehrnia]

Hi,

these routes in different area's would be independent from eachother. in this case, i think each device will prefer the route (10.100.0.0) which is in its own area (Intra-Area), and the ABR would simply disregards the 10.100.0.0 route comming from Area 10 due to a greater metric,

it is not a stable and legit form of design anyway.

Hope it Helps,

Soroush.

[Mohammed Islam]

Thanks.  Yes there are multiple area (approx 15).  I am just concerned as to having the same network advertising in different OSPF areas whether that would case any forms of asymetric routing, routing loops etc. 

Would it be metric/cost that it decides on?

[smehrnia]

Correct me if im wrong, but i'd say in such scenario when generating LSA-3, ABR would advertise the route with the lower metric/cost into other Area's, which i believe normally it is the one that is connected to the HEADEND ABR itself.

but in case, a same network (different Area's) comes from different ABRs then there could be unpredictable results in the Metric/Cost selection process.

Hope it Helps,

Soroush.

[Mohammed Islam]

Correct regarding the ABR and network advertisements.

So I assume this is not highly advised?!

At present:-

MAIN SITE:-

OSPF 1 process

Switch

Network 10.100.0.0 area 0 > advertise > to other OSPF areas including AREA 1 (type 3 lsa)

FAR END

OSPF 1 process

Switch

Network 10.100.0.0 area 1 > advertise > back to MAIN site (type 3 lsa)

Network 11.0.0.0 area 1 > advisetise > back to MAIN site (type 3 lsa)

Although I can manupilate costings, I think there may be asymetrical routing issues or some form of loops.

To make it a bit more complex, there are other VPN sites which are in the same area 1 in a ring formation.  They generally talk to that 10.100.0.0 network on the main site.  I think by adding this network to the 'FAR END' as mentioned above it may also disrupt their routing ability in a 'normal' situation.

[smehrnia]

this Highly not recommended at all...! you should avoid such design.

regarding ur question, it should be lab-ed to see the results, i cant exactly tell u that, but this might do that to ur network.

HTH

plz Rate helpful posts, for later reference of other users.

Soroush.