Frame Relay Question


I have a question please :

What is the difference between point-to-point & multi-point subinterfaces, why should we multi interfaces instead of physical lines?

please no web site ressources or pdf books, i need a simple explaination from your own experience, the simpler answer is the more i can get it easily

thank you
Who is Participating?
ngmarowaConnect With a Mentor Commented:
1. Multipoint interface reduce the number of cables required
2. multipoint interfaces reduce the number of interfaces required for example if you need 4 links a router will need only one physical interface to handle them instead of having 4 physical interfaces. You will need expensive equipment to achieve this as most routers do not have room for many physical interfaces.
3. The logical interfaces work just as well as the physical interfaces since you can assign different IP to each or even different configurations

Kindly explain why you would need physical interfaces instead of virtual interfaces? I believe there are instances where physical interfaces are more desirable than logical interfaces but these are rare.
that's not as simple a question as one might think it is, however, I will try my best! One of the differences are how each type (P2P and P2MP) handles inverse arp and what type of network they represent.

point to point (one to one) links are treated like point to point links (only 2 devices on that link);basically, there is no need to map or inverse arp. Due to the nature of a point to point link, the protocol just assumes that whatever IP you're trying to reach is on the other end of that link, so the IP packet is encapsulated in a HDLC packet and sent out on the link.

Point-to-multipoint links are considered as a multi-access medium. (one to many [Hub to Spokes])

If all your remote edge devices need to be on the same logical subnet, you use multipoint, if all your remote edge devices can be addressed with individual subnets, you use point to point.

Hope this helps a bit

david875Author Commented:
@ngmarowa I think that you have a simple way to explain, i asked you about the use of physical interfaces instead of virtual ones just to have a general idea and study the frame relay in every side that's it, of course i would use logical interfaces, but when you said the Multi point interface reduce the number of cables required and explain with the exemple i have understand that multipoint plays like if i have many physical lines in the same Serial cable. You still didn't give me an  idea about the point-to-point subinterfaces' use.
Can you tell me when can I use Multipoint interfaces and point-to-point and why?

@rfc1180: you said : there is no need to map or inverse arp to use point-to-point and so the IP packet is encapsulated in a HDLC packet and sent out on the link... This looks to me like if i'm configuring a simple Serial cable between 2 routers without puting Frame Relay at all !!! What about Frame Relay encapsulation?
And what you mean with  :Point-to-multipoint, can you detail this?
You also mentionned that in individual subnets i can use point-to-point, so how about a routing protocol like RIP or OSPF to be configured in other routers ?

Thank you
Get expert help—faster!

Need expert help—fast? Use the Help Bell for personalized assistance getting answers to your important questions.

Multipoint interfaces are used when you want to connect 2 or more sites using one physical interface. Have used it for a Head Ofiice - Branch Offices configuration. Its like a hub and spoke setup. The branch offices will see it as a point to point. Only the Head office will have a multipoint interface. A more complicated approach would be a mesh topology where the branch offices also have multipoint interface to 2 or more point.

The bottom line is what you want to achieve and what resources are available. An example you want to link 3 sites, A, B and C. There are 2 possibilities:

1 A has a link to B and B has a link to C but  A and C do not have a direct link
2. All 3 sites have direct and independent links to each other

For scenario 1 you can have A link to C through B but the link will seem like a direct link until either the link between A and B or B and C is down. A and B will be linked with a point to point link. The same link is used for B to C. So B can use a multipoint interface.

Scenario 2 is open to a number of options. The bottom line is Technology, resources and infrastructure determines when you can use multipoint or point to point. Also point to point is used to connect 2 sites/points. When you have more than 2 then multipoint or multiple point to point links maybe required.
>there is no need to map or inverse arp to use point-to-point and so the IP packet is encapsulated in a HDLC packet and sent out on the link

"Sorry meant to say frame-relay packet not hdlc packet";it is not required to enable or disable Inverse ARP, because there is only a single remote destination on a point-to-point PVC and discovery is not necessary.

>You also mentionned that in individual subnets i can use point-to-point, so how about a routing protocol like RIP or OSPF to be configured in other routers ?

Routing protocols: this is a totally different topic. OSPF has 5 different types for FR-NBMA and how they react with point to point and point to multipoin. Some of the topics related aret:

    * OSPF treats Point-to-Multipoint networks as a series of point-to-point links, mirroring the layer 2 topology.
    * Point-to-Multipoint networks don’t have DR/BDR relationships.
    * Point-to-Multipoint networks advertise /32 routes for all the frame-realy endpoints.
    * Point-to-Multipoint networks have Hello and Dead intervals of 30 seconds and 120 seconds respectively.
    * Point-to-Multipoint non-broadcast networks are very similar to point-to-multipoint networks except that they work over layer 2 topologies that have no broadcast capability
    * Point-to-Multipoint non-broadcast networks allow you to define cost on a per neighbor basis.

There is no real simple answer to your question; it really depends on your network topology, routing infrastructure, etc. There are many designs that you can choose from based on how you want to deploy the network.

>please no web site ressources or pdf books, i need a simple explanation from your own experience

yet we have given you a simple explanation, but you still have questions; the only way you are going to understand it is from many resources on the web and possibly tech books on the subject. You are asking a question that can lead to a full chapter and more of explanations and no simple answer can honestly be given.


david875Author Commented:
rfc1180: Thank you for your effort, I know this is not simple and of course this can lead to read books i just wanted something easier to explain it, web sites and tutorials seems to be complicated and they don't explain anything from the scratch.

Anyway, I've seen a scenario recently about configuring Frame relay here is the Scenario of a Hub & Spokes:

                                                =======     ||   Frame   ||   =======
                  Router3      =====  \\                ||   Relay    ||               //=====  Router 2            

Router 1 is the Hub
Router 2 & 3 are the Spokes

I've seen some videos where R1 is configured as Point-to-point frame relay dlci with R2 & R3

Router2 need to go through R1 to have a communication with Router3,

Can you explain me why should we configure point-to-point and multipoint, is it about choice ? or about designing solutions? I think this is not something we guess and then say :" well i think i will configure point to point because i love it !!!" right?

If we have to configure point-to-point, what will happened if we configure multi point in this case?

Thank you again

Don JohnstonConnect With a Mentor InstructorCommented:
Let me take a shot at this...

We're talking about point-to-point and multipoint SUB-interfaces.

Without sub-interfaces, all PVC's terminate (for lack of a better word) on the physical interface. So if you've got 30 PVC's, they're all part of the same IP network. Which (by extenstion) all the routers at the other ends of those PVC's will have their serial interfaces on the same network.

Scenario 1

In your diagram, we could do this without using sub-interfaces.

R1 s0/0
R2 s0/0
R3 s0/0

Router 1 will have a PVC going to Router 2 and another one going to Router 3.

Now the problem with this is that if we use a routing protocol like RIP. Router 3 sends out an update which arrives at Router 1 on it's serial 0/0 interface (since there's no PVC from R2 to R3, it won't go to R3). But split horizons prevent R1 from advertising the route that it learned from R3.

That's why we have point-to-point sub-interfaces. R1 (and only router 1) would create two sub-interfaces. One for each PVC. The PVC to R3 would be s0/0.3 and the PVC to R2 would be s0/0.2. Now when an update from R3 arrives at R1, it's coming in on s0/0.3 so we're allowed to send it out s0/0.2.

Scenario 2

But because each interface has to have an IP address, we now have two addresses associated with s0/0. So we end up with something like this:

R1 s0/0.2
R2 s0/0
R3 s0/0

So now instead of all the routers being on the same network, they're now on two different networks. This could be a bit of a pain with a large number of PVC's.

On to the multi-point sub-interface.

You could have done Scenario 1 with a multi-point sub-interface on Router 1. But there's absolutely no reason to so. There would be no difference in the behavior (just a lot more configuration commands).

So where would you use a multi-point sub-interface?

Scenario 3

If you had a bunch of PVC's coming in to one site. And you wanted some of those PVC's on the same network and the rest treated as separate. You would create point-to-point sub-interfaces for the PVC's that you wanted as separate and you would create one (or more) multi-point sub-interfaces for the PVC's that you wanted to treat as being on the same IP network.

Hope this helps.

rfc1180Connect With a Mentor Commented:
>Router2 need to go through R1 to have a communication with Router3,
unless you specify the mappings to and from R2 and R3.

>think this is not something we guess and then say :" well i think i will configure point to point because i love it !!!" right?

Correct, you will use either type based on your requirements (what do you want to do and how are you going to do it).

A point-to-point subinterface establishes 1 permanent virtual circuit  to another physical interface or subinterface on a remote router placing them in their own subnet. In a point-to-point network environment, each subinterface is acting like a point-to-point interface. Typically, these are in a separate subnets for each point-to-point virtual circuit. Therefore, routing update traffic is not subject to the split horizon rule.

A multipoint subinterface establishes multiple permanent virtual circuits to multiple physical interfaces or subinterfaces on remote routers and are all in the same subnet. The subinterface acts like an NBMA Frame Relay interface (Non-Broadcast in nature), so routing update traffic is subject to the split horizon rule. normally, all multipoint VCs belongs to the same subnet.

Do all your sites require connectivity to one another?
Do you have adequate bandwidth at the hub for spoke to spoke traffic?
Will you be running routing protocols, and if so, different routing protocols have different requirements on the type of network they connect to (Broadcast vs. NBMA [Non Broadcast Multi-Access).
Some routing protocols work better over point to point links, some require full mesh designed that requires you to use point to multipoint.
Are all your VCs the same bandwidth?

To dive into each of these is not easy and is very lengthy:


Question has a verified solution.

Are you are experiencing a similar issue? Get a personalized answer when you ask a related question.

Have a better answer? Share it in a comment.

All Courses

From novice to tech pro — start learning today.