Routes from one standard area to other standard area is not passing via backbone ie, ABR

[Vignesh M]

As per my knowledge routes of one standard area will be shared with other standard area using backbone area. Eg: Area 1 ->Area 0 -> Area 2.

I have simulated a scenario and found a strange behaviour by OSPF. Please find the attached topology.

Advertised all the interfaces of all the router in the respective areas as per te topology.

Scenario 1:

Now traceroute 6.0.0.1(loopback interface in R6) from R5, The transit path is R5->R2(ABR)->R1->R3(ABR)->R6

In this scenario, the protocol works as expected. I mean, to reach a network from standard area to another standard area it has to go via backnone area(ABR's).

Scenario 2:

Shutdown the link s1/0 in R5. Now traceroute 6.0.0.1(loopback interface in R6) from R5. The path is R5->R4->R6.

In this scenario, the protocol is misbehaving. I mean to reach 6.0.0.1 which is in R6(Area 2) from R5(Area1), I have to go through ABR. It should be like R5->R2(ABR)->R1->R3(ABR)->R6. Please clarify this strange behaviour.

[Joseph W. Doherty]

Disclaimer

The Author of this posting offers the information contained within this posting without consideration and with the reader's understanding that there's no implied or expressed suitability or fitness for any purpose. Information provided is for informational purposes only and should not be construed as rendering professional advice of any kind. Usage of this posting's information is solely at reader's own risk.

Liability Disclaimer

In no event shall Author be liable for any damages whatsoever (including, without limitation, damages for loss of use, data or profit) arising out of the use or inability to use the posting's information even if Author has been advised of the possibility of such damage.

Posting

No, it's not strange behavior.

R4 "knows" both area 1 and area 2 topology, so when traffic from R5 hits it, actually on the way to R5 via R2, R4 routes it directly to R5.  You should see the same result if instead of shutting s1/0 on R5 you costed it such that s1/1 would be the better path to R2.

Although R4 seems to be behaving as an ABR, it's not, because it doesn't share the area 1 topology with area 2 or the converse.

To preclude R4 from doing what you're seeing, you would need to have the two OSPF areas in different VRFs.

The same would also happen if instead of R4 running OSPF on both sides, one side was OSPF and the other side was a different routing protocol.  Again, the key point is R4's routing table has both sides routes within it.

BTW, packets from R6, with R5's s1/0 shut, should flow R3,R1,R2,R4 (assuming R3 has a lower cost to it via the direct link).

[Rolf Fischer]

I'd just like to add a little detail to what Joseph has already explained.

In scenario 2, R5 indeed calulates the path to R6's loopback through the backbone area, as you can see from the cost of that routes. But routing is always a local decision and R4 (the next-hop router in the path) sees all networks from area 1 and area 2 as intra-area, so it can forward this traffic directly.

HTH

Rolf

[Joseph W. Doherty]

Disclaimer

The Author of this posting offers the information contained within this posting without consideration and with the reader's understanding that there's no implied or expressed suitability or fitness for any purpose. Information provided is for informational purposes only and should not be construed as rendering professional advice of any kind. Usage of this posting's information is solely at reader's own risk.

Liability Disclaimer

In no event shall Author be liable for any damages whatsoever (including, without limitation, damages for loss of use, data or profit) arising out of the use or inability to use the posting's information even if Author has been advised of the possibility of such damage.

Posting

Rolf's makes a great point, but although he's correct about routing is a local decision, we must also account for what's in the router's routing table that pertains to the received packet.  Again, if R4 was using VRFs, you could have multiple route tables, whose contents, by default, are not shared.  Or, if you had a router that supports virtual routers, you could have multiple logical routers.  In such cases, R4 probably would not pass packets to the physical interface with a more direct connection (and it would behave as you expected).

I suspect your R4 uses a single ospf process, but if you want to try another variation, use two OSPF processes.  This would let you have a different RID for R2 in area1 and area2, yet your results will be the same without having a VRF.

[Vignesh M]

Thanks for the replies. Yes Joe. My R4 is having single ospf PID. 

Now I want to reitrate the point that, though the IA routes learned from ABR, if a router have route to a perticular destination IP, then it need not to go via ABR.

[Joseph W. Doherty]

Disclaimer

The Author of this posting offers the information contained within this posting without consideration and with the reader's understanding that there's no implied or expressed suitability or fitness for any purpose. Information provided is for informational purposes only and should not be construed as rendering professional advice of any kind. Usage of this posting's information is solely at reader's own risk.

Liability Disclaimer

In no event shall Author be liable for any damages whatsoever (including, without limitation, damages for loss of use, data or profit) arising out of the use or inability to use the posting's information even if Author has been advised of the possibility of such damage.

Posting

"Now I want to reitrate the point that, though the IA routes learned from ABR, if a router have route to a perticular destination IP, then it need not to go via ABR."

In this case, no it doesn't, but again, that's because R4 also "knows" a more direct path than then ABR.

If there was another area, that R4 didn't have a connection to, it would route traffic via the ABR, as expected.

Also keep in mind, unlike the ABR, it doesn't inform the rest of either OSPF area about the other OSPF area's routes.  Again, take special note of return traffic.

[Vignesh M]

I suspect your R4 uses a single ospf process, but if you want to try another variation, use two OSPF processes.  This would let you have a different RID for R2 in area1 and area2, yet your results will be the same without having a VRF.

---------------------

Hi Joe,

I have increased path cost of the s1/0 in R5. Now to reach 6.0.0.1 from R5 , path is R5->R4->R6.

I configured R4 with different PID, and isolated A2 and A1 in R4. Eventhouth PID's are different, routing table shows R6 to reach 6.0.0.1 as it is shortest distance.

Hence options to isolate is using VRF and next one is someother routing protocols.

Regards,

Vignesh