OSPF, like any other routing protocols, can agree or decline neighborship if their requirements are not met. Here are the adjacency requirements in order to form neighborship with OSPF:

1. same area ID
2. same subnet
3. same authentication type and authentication password
4. same hello and dead timers configuration
5. same area type

AREA ID

You configure Hyla s2/0 interface in area 1 and you configure Anaxyrus s2/1 interface in area 0. Do you think they are going to be neighbors? No. The first thing that OSPF checks is to form adjacencies with routers in the same area. It will be a mismatch if they are not on the same area.

Bringing back my topology on my previous post:

Anopheles(config)#router ospf 2
Anopheles(config-router)#network 192.168.12.0 0.0.0.255 area 0

Culex(config)#router ospf 1
Culex(config-router)#no network 192.168.12.0 0.0.0.255 area 0
Culex(config-router)#network 192.168.12.0 0.0.0.255 area 10
*Oct 18 23:54:34.411: %OSPF-4-ERRRCV: Received invalid packet: mismatch area ID, from backbone area must be virtual-link but not found from 192.168.12.2, Serial2/0

An invalid packet has seen indicating a mismatch area ID.

SUBNET MASK

A subnet mask is part of a Hello packet and one of the criteria that OSPF looks to form adjacencies. If you configure the two links in a different subnet, it will show a mismatch in hello packet parameters. This indicates that both links have different subnet masks. There is somehow a debate about this because this one should not be a mandatory requirement for OSPF adjacency. Subnet mask in point-to-point links is ignored and formed neighborship even connected links don’t have the same subnet mask. I will show you my lab example in my future posts.

AUTHENTICATION TYPE/AUTHENTICATION PASSWORD

As the name explicitly states, it is all about authentication configuration on OSPF. You must sure that if you have configured authentication on one link, you should have the same authentication configured on the remote link as well. And the authentication type should be the same: 0 – no password, 1 – plain text, and 2 – MD5 password. Lastly, the authentication key should be the same as well. I have a separated full-blown post regarding OSPF authentication.

HELLO AND DEAD TIMERS

EIGRP doesn’t care about its hello and dead timers but OSPF does. Hello and dead timers are OSPF Hello parameters and they must match or else adjacencies will not be formed. These two parameters are very crucial in determining if a neighbor still exists or not. A hello interval indicates how often the hello packet is sent while the dead timer indicates how long the neighbor router should wait for hello packets before declaring its link dead. The default hello interval for broadcast and point-to-point links is 10 seconds while dead timer is 40 seconds. And for the NBMA, point-to-multipoint, and point-to-multipoint NBMA hello timer default is 30 seconds while dead timer is 120 seconds. Did you notice that the dead timer is always four times the value of hello timer?

AREA TYPE

This is where OSPF area types like Stub, Totally Stubby, NSSA, and Totally NSSA come into play. Looking back on my previous post regarding area types, if you want to configure a certain area as Stub, Totally Stubby, NSSA, and Totally NSSA all the routers in this area should be in the same area type as well or else no adjacencies will be formed.

OSPF NETWORK TYPE
There are different OSPF network types: Broadcast, Non-Broadcast, Point-to-point, and Point-to-multipoint. Network type is not usually tagged as one of the mandatory requirements to form neighborship with OSPF. However, incompatible network types can’t form adjacency to each other. These network types can be further classified as DRs and non-DR types. Combining a DR and non-DR will not form OSPF adjacencies.

I will use the same lab in my previous post. I just changed the configuration on the Anopheles router and put the “no-summary” command.

Anopheles#conf t
Enter configuration commands, one per line. End with CNTL/Z.
Anopheles(config)#router ospf 2
Anopheles(config-router)#area 20 nssa no-summ
Anopheles(config-router)#area 20 nssa no-summary
Anopheles(config-router)#end

Aedes(config)#router ospf 3
Aedes(config-router)#no area 20 stub
Aedes(config-router)#
Aedes(config-router)#area 20 nssa

It would be better to check the differences between NSSA and Totally NSSA using the before and after the output of the NSSA configuration.

Unlike NSSA, Totally NSSA once configured on the ABR, will automatically inject the default route to advertise Type 3 LSAs. So, there is no need to explicitly configure it. Am I repeating the same thing?

Just like what happened in the Totally Stubby Area database, Type 3 LSAs are restricted and all that is left is the Type 1 LSA, Type 7 LSA and the default route advertised by 192.168.2.1.

My posts regarding OSPF Area types are not yet done. If you will take a closer look, there are many things that I haven’t touched yet or clarified yet. Like if you check the database output, why some LSAs are missing even though it is allowed in this kind of area? This is going to be another discussion in my future posts.

According to Cisco, OSPF area is a collection of logical OSPF networks (routers and links) that have the same area identification. Depending upon the design of your network, your routers can be part of a single area or multiple areas. Why do we have to complex things and put the router in different areas? Actually, there are many reasons, and one of the many reasons is that it reduces the size of OSPF database when OSPF network is divided into different areas. Think of it this way, dividing the routers into different areas, reduces the size of the database, reduces the frequency of SPF calculation and smaller routing table. Thus, fewer requirements on router memory and CPU. I am not saying that SPF calculation is exhausting the CPU and router memory but the sending and flooding of the new topology information does.

OSPF AREA TYPES

No matter what your OSPF design in your network, all areas you have created should be connected to the backbone area. Backbone area or Area 0 (0.0.0.0) is like the meeting points of all these non-backbone areas. So, it is not just an option or recommendation but a must in every OSPF network design. Although there is an option called virtual link, it is not a good network design but just a remedy to a broken OSPF design.

Now, not all LSA Types are allowed in different areas. Some LSA Types are restricted and depend upon which area type is configured. This further reduces the link state database and routing tables. Injecting too many external routes are also memory intensive. However, all LSA types are allowed in the Normal area. An area which is not configured as Stub, Totally Stub, NSSA, NSSA Totally Stub area is called Standard/Normal Area. We can say that the backbone area is behaving like a Normal /Standard Area as LSA Type 1-5 are not being restricted in this area. When the network is divided into different area numbers (non-zero), an ABR is used to connect the Standard area to the Backbone Area.

On my next post, I will introduce some of the area types which have their own rules in restricting LSA Types.