OSPF Between Areas

up2thetime

Hey guys... I'm practicing a lab scenario, and came across something confusing.

R2 is an ABR and has one interface in Area0, Area1, and Area3.
Here is the topology:
R2(Area1 Interface)

---

R3(Area1 Interface)
R2(Area3 Interface)

---

R8(Area3 Interface)
R3(Area3 Interface)

---

R8(Area3 Interface)
R1(Area1 Interface)

---

R3(Area1 Interface)


R1 is a router in Area 1. When it wants to reach an Area3 subnet it takes the following path:
R1 - R3 - R2 - R8. It must go through R2 since R2 connects to the backbone area.

Here is the part I don't understand:
When R8 (Area3) wants to reach R1 (Area1), the traffic goes straight through R3.
Why isn't R2 used? I thought all traffic must flow through a ABR which connects to Area 0.
In this case, traffic is flowing straight from Area1 to Area 3. No virtual link is in place.

Thanks.

Update:
Since R3 has interfaces in Area 1 and Area 3, it looks like R3 is advertising Area 3 routes into Area 1.
When I trace from Area 1, it goes straight though R3, and into Area 3. However when coming from Area 3, the traffic does not go directly through R3- it goes through R2 (which is connected to Area 0).

I went over again how the LSAs are flooded.
I guess R8 does flood the Router LSA over to R2, which in turn would inject a Summary LSA over to R3. However R2 has a directly connected link between itself and R8, so does not install the ospf route- instead it uses its connected route.

pogue

Frankly, without both a network diagram, configs, and sho ip routes, I am totally guessing here...

I have never seen a lab like what you described. I sketched it out, and Router2 and Router3 are both ABRs between areas 1 and 3, with Router 2 pulling additional duty as the link to the backbone. I don't think this is a common configuration at all.

I don't think areas 1 and 3 would normally be connected like this. A more common topology would be for area 1 and 3 to be at opposite ends of the internet, and both would route through the backbone to get to each other, easy-peasy.

By connecting areas 1 and 3 together, I believe R2 might be advertising an interarea route for R1 network to R8, but to get back R3 may be advertising an intraarea route to get back to R1 that is seen as more favorable than the interarea link back.

It seems like this might be a lab you put together yourself. If so, you might want to put R3 in either area 1 or area 3, disconnecting it from the second router.

Remember, OSPF is supposed to be hierarchical. Packets are supposed to flow from area 1 to area 0 to area 3, and vise versa. When you add a redundant link on R3 to area 3, R3 may inject aleternate paths to area 3 that can cause asymmetric routing.

Russ

up2thetime

pogue wrote: »

Frankly, without both a network diagram, configs, and sho ip routes, I am totally guessing here...

I have never seen a lab like what you described. I sketched it out, and Router2 and Router3 are both ABRs between areas 1 and 3, with Router 2 pulling additional duty as the link to the backbone. I don't think this is a common configuration at all.

I don't think areas 1 and 3 would normally be connected like this. A more common topology would be for area 1 and 3 to be at opposite ends of the internet, and both would route through the backbone to get to each other, easy-peasy.

By connecting areas 1 and 3 together, I believe R2 might be advertising an interarea route for R1 network to R8, but to get back R3 may be advertising an intraarea route to get back to R1 that is seen as more favorable than the interarea link back.

It seems like this might be a lab you put together yourself. If so, you might want to put R3 in either area 1 or area 3, disconnecting it from the second router.

Remember, OSPF is supposed to be hierarchical. Packets are supposed to flow from area 1 to area 0 to area 3, and vise versa. When you add a redundant link on R3 to area 3, R3 may inject aleternate paths to area 3 that can cause asymmetric routing.

Russ

Thanks Russ!

Yea it was just something I made up for testing.

When traffic flows back from R1 to R8, R3 realizes it has connected routes to the destination. I guess thats why the OSPF IA routes never make it to the routing table (however R2 is actually sending the Summary LSAs). To test this, I shut down the link between R8 and R3 to cause an SPF recalculation. As expected, the connected routes are no longer in R3s routing table, and all return traffic flows through R2.

Thanks again!

pogue

No problem man.. Good luck with your studies!

Russ