NSSA purpose

Routing & Switching OSPF
1

Hello,

In a NSSA, type 5 LSA is replaced by Type 7, that means the amount of LSA flood remain same. So what are we achieving in terms of minimising LSA flooding, which was the original reason for creating a stub area. Also what makes NSSA special than a standrad area apart from the fact that they use two diffrent LSA types for the same purpose.

2

Hello Don.

I think that the best definition of an NSSA is that it’s an area that allows redistribution from routers in the local area but not from other areas.

I will use a diagram by Ed (PracticalNetworking on YouTube).

image

We have 4 areas where in Area 22 and 44, there are routers (R5 and R7) redistributing routes from a foreign domain. If we count all the redistributed subnets together, we would end up with 200 type 5 LSAs (100 from R5 and 100 from R7).

Pay attention to Area 44. Right now, we have 200 redistributed subnets. We could configure Area 44 as a Stub area which would effectively block out any type 5 LSAs.

However, this would create a problem for us. The 100 external subnets R7 redistributes into Area 44 would never make it because a Stub area blocks Type 4/5 LSAs thus we would be without connectivity to them.

The best option here would be to configure Area 44 as an NSSA area. An NSSA area blocks out any Type 4/5 LSAs from foreign areas but it allows redistribution from the local area by using a Type 7 LSA. Why Type 7 LSA? Well since we are blocking Type 4/5 LSAs, we need a way around this to allow redistribution from the local area.

The end result would look like this:

image

The 100 Type 5 LSAs that came before from a foreign area have now been blocked but we still allow redistribution from the local area by using a type 7 LSA. So we’ve effectively minimised our LSDB and routing tables while also allowing redistribution from the local area.

If you have any further questions, please let me know.

David

3

Hello David

Thanks for responding so comprehensively to this post! It’s much appreciated! Just to let you know, Rene has gone in and replaced the copyrighted material that you linked to and showed. Not to worry, you shared it in good faith, however, it’s been modified just to be safe. I believe that the replaced content still describes the explanation correctly.

In any case, @Donjosephweb if you need any further clarifications, let us know.

Laz

1 Like
4

Thanks David for the explanation and i appreciate your emphasis on each points where i had doubts. However, let me state/ ask some points for which I request for more clarification.

  1. Is my understanding correct that, a stub area doesnt allow redistribution of LSA type 5. Therefore ASBR blocks it, thus local routers in stub doesnt learn them(100 routes in case of Area 44 here). But in case of NSSA each local routers learn routes locally as type 7 and the ABR R4 will pass them to area 0 as type 5?

  2. I do not still understand how we are minimising LSDB because wouldnt 100 routes as either LSA 5 or LSA 7 occupy the same space in LSDB of Area 44?

5

Hello Don

There are a couple of clarifications needed here. Here’s an example of a topology where a stub area can be configured:

image
image712×534 44.2 KB

R1 is an ASBR and R2 is an ABR.

Now, if the above were configured as a normal OSPF area, then both R1 and R2 would advertise Type 5 LSAs into area 2 which will contain information about external routes. R1 would learn them directly from the RIP router which is outside of the OSPF domain. R2 would learn them from other areas of the OSPF topology where an ASBR may exist.

Now stub areas are designed to reduce the OSPF routing table size and simplify the management of the OSPF network. In a stub area, LSA Type 5 (External LSAs), which are used to distribute routes external to the OSPF autonomous system, are not allowed. So if the above were configured as a stub area, then no Type 5 LSAs would be allowed. They are not allowed to be advertised by either ABR or ASBR. Instead, the ABR will inject a default route into the area for all traffic.

But there’s a problem. R1 has learned about routes from the RIP router. If area 2 is a stub area, R2 can’t inject any of the routes it has learned from the RIP router.

Therefore, we create an exception to the rule of Type 5 LSAs. With an NSSA, we can use a Type 7 LSA so R1, an ASBR, can inject the routes it learns from the RIP router into the NSSA.

Now we can answer your second question:

A Type 7 LSA can only be generated by an ASBR, and not an ABR.

  • So in a normal OSPF area, both ABRs and ASBRs can generate type 5 LSAs and inject them into the area. So external routes enter the area from both ABRs and ASBRs.
  • In a stub area, neither ABR nor ASBR can generate type 5 LSAs. So external routes cannot enter the area from either ABRs nor ASBRs.
  • In an NSSA area, only the ASBR can inject a type 7 LSA while an ABR cannot, so external routes can enter the area ONLY from ASBRs and not ABRs.

So in an NSSA, we are reducing the size of the LSDB by not accepting any external routes from the ABR, but we are accepting external routes from the ASBR. Does that make sense?

I hope this has been helpful!

Laz

6

This is perfect Lazarus. Thanks as always :smile:

1 Like