OSPF w/ Point-to-point vs Point-to-multipoint

e24ohm

Folks:
I am currently studying for the CCNP BSCI and I'm using the following books at the moment: CCNP BSCI Portable Command Guide by Scott Empson, and CCNP BSCI Official Exam Certification Guide 4th ed. by Brent D. Stewart and Clare Gough.

Right now I am in Chapter 6 of the Official Exam Certification Guide and having a hard time getting to understand OSPF deployed on Point-to-point and/or multi-point. The books says that the easiest way and straightforward way to deploy OSPF in a mesh environment is to use Point-to-Point circuit. Would this be configured as a subinterface?

[stewart, pg. 172] “The third option is to treat each PVC as a separate point-to-point circuit. This is the easiest method to document and to troubleshoot, but create a more involved configuration. This is a very common way to handle this situation.”

Then it says [stewart, pg. 172] “The final option is to create two subinterfaces on Rotuer A. A multipoint interface would use NBMA or broadcast OSPF network types on the interfaces in the A-C-D full mesh, while a point-to-point subinterface would be used on Router A to connect to Router B.”

I am confused, do I create subinterfaces in both cases? I forget, what is the maximum amount of subinterfaces can a physical port have? Doesn't the creation of subinterfaces take more memory?

Thank you,
E

cisco_trooper

It is best practice to configure your point-to-point on subinterfaces. Multipoint interfaces are configured on the physical interface. On modern routers the use of a subinterface really isn't going to have much effect on your device. There is plenty of memory available.

I found Routing TCP/IP Volume I, Second Edition to be a fantastic resource when it came to the OSPF topics. The BSCI Exam Guide leaves ALOT to be desired in my opinion.

Routing TCP/IP, Volume 1, 2nd Edition - $72.00

cisco_trooper

Oh, and the best way to get a handle on these particular topics is to lab them up. Then lab them up again. And then do it again. OSPF over NBMA was one of the more difficult topics for me to grasp for WHATEVER reason.

EdTheLad

e24ohm wrote: »

Folks:

Right now I am in Chapter 6 of the Official Exam Certification Guide and having a hard time getting to understand OSPF deployed on Point-to-point and/or multi-point. The books says that the easiest way and straightforward way to deploy OSPF in a mesh environment is to use Point-to-Point circuit. Would this be configured as a subinterface?

[stewart, pg. 172] “The third option is to treat each PVC as a separate point-to-point circuit. This is the easiest method to document and to troubleshoot, but create a more involved configuration. This is a very common way to handle this situation.”

I think you are confused between your L2 and L3, are you asking about multipoint interfaces or ospf point-to-multipoint.

A serial interface by default is classified as a multipoint interface, this means more than one virtual circuit can terminate on it.You can create a subinterface, this subinterface can be configured as either point-to-point or multipoint.
A point-to-point subinterface means there will only be one remote host and both hosts will share the same subnet, typically a /30.
A multipoint means you can have many VCs terminating on the port, these VCs will connect to one or more remote neighbors, they will all share the same subnet, so if you had 4 remote neighbors typically you would use a /29 mask.

Now depending on how you have previously setup the interface i.e. ptp or ptmp, the ospf network type will change.By default a multipoint physical interface will be nmba, a ptp subinterface will be ptp(if i remember correctly).
Depending on how devices are phyically connected you can modify the ospf network type.The problem with nmba is that all neighbors have to be manually assigned using the neighbor command,also a DR needs to be elected.Ospf network type broadcast can be used in a full mesh environment, the l2 must have broadcast capability, this means the neighbors can be discovered dynamically so no need for the network statements, but a DR is still elected.Another method is "point-to-multipoint" this creates a mesh of ptp connections, allowing for dynamic neighbor discovery and no need for a DR.
Then there is the final type "point-to-multipoint nonbroadcast" this is the same as ptmp excpte since there is no broadcast capability at L2 the neighbor command is required to statically setup neighbors.

acidsatyr

e24ohm wrote: »

Folks:
I am currently studying for the CCNP BSCI and I'm using the following books at the moment: CCNP BSCI Portable Command Guide by Scott Empson, and CCNP BSCI Official Exam Certification Guide 4th ed. by Brent D. Stewart and Clare Gough.

Right now I am in Chapter 6 of the Official Exam Certification Guide and having a hard time getting to understand OSPF deployed on Point-to-point and/or multi-point. The books says that the easiest way and straightforward way to deploy OSPF in a mesh environment is to use Point-to-Point circuit. Would this be configured as a subinterface?

[stewart, pg. 172] “The third option is to treat each PVC as a separate point-to-point circuit. This is the easiest method to document and to troubleshoot, but create a more involved configuration. This is a very common way to handle this situation.”

Then it says [stewart, pg. 172] “The final option is to create two subinterfaces on Rotuer A. A multipoint interface would use NBMA or broadcast OSPF network types on the interfaces in the A-C-D full mesh, while a point-to-point subinterface would be used on Router A to connect to Router B.”

I am confused, do I create subinterfaces in both cases? I forget, what is the maximum amount of subinterfaces can a physical port have? Doesn't the creation of subinterfaces take more memory?

Thank you,
E

You are mixing Frame-Relay configuration (l2) from OSPF over Frame-Relay configuration (l3).
Look into how Frame-Relay is configured to work, no matter what routing protocol your implementing over it.
There are 5 ways to configure OSPF over FR, and they all have to do with:
-do neighbors discover each other ? (NBMA vs Broadcast),
-is there DR/BRD (multipoint don't have this, since it's viewed as a collection of point2point)
-timers (30/120 vs 10/40 sec).

Look at how EIGRP is implemented over FR.

gl.

kryolla

Also when doing NBMA network type ask yourself why I am putting a neighbor statement, why does the DR have to be the hub, why do I need to set the priority to 0 for the spokes, why do I need a FR map statment for spoke to spoke comm. You need to know the underlying layer 2 in order to understand the different network types for OSPF. Also certain network type isnt only bound to ethernet or FR i.e broadcast is only for ethernet.