inverse ARP

Hi guys For Frame Relay networks, DLCIs (Data Link Connection Identifiers are used) for establishing Virtual cirucits for communication. The inverse ARP helps in filling the table which maps IP addresses and DLCI used for communication. For filling the tables, ARP messages are used over the Virtual circuits for getting the IP addresses of the terminals on the remote side. I am consfused as how is a Virtual cirsuit established in the first plae if one of the Comps does not know the IP address of the remote comp. Please clarify. thank you Shantanu

As i understand it, Frame relay operates at layer-2 so it does not need an IP address to establish it's virtual circuit. I think it finds out the remote dlci and that is all it needs to establish.

somebody correct me if I'm wrong.

Also make sure you know frame relay good, I got alot of frame relay on the ccna exam.

genki napisa=B3(a):

Yes FR is a Layer 2 protocol. LMI at the switch side informs DTE about active Virtual Circuits DTE use Inverse Arp to send its IP address via VC - the same happens=20 from the DTE on the other side. After a while, both DTE have correct mapping VC - IP address. You can disable Inverse Arp and manually configure appropriate mappings.

------------------------ Ern

Yes FR is a Layer 2 protocol. LMI at the switch side informs DTE about active Virtual Circuits DTE use Inverse Arp to send its IP address via VC - the same happens from the DTE on the other side. After a while, both DTE have correct mapping VC - IP address. You can disable Inverse Arp and manually configure appropriate mappings.

------------------------ Ern

Well this confuses me also. Normally you, as administrator, place IP addresses on interfaces. So how does this Inverse ARP action know whats an appropriate IP address for the DTE (router) interface? Does this rely on some kind of DHCP server?

Obviously I'm missing something here, Please explain a litte more.

Thanks!

The mechanism by which inverse ARP works and the motivation for the protocol is well explained in the relevant RFC (RFC 1293).

Take a look at

Cisco da Gama

Well, thanks for the link I'm still not getting it.

I read for example this:

7.2. Protocol Operation Within Frame Relay

One case where Inverse ARP can be used is when a new virtual circuit is signalled. The Frame Relay station may format an InARP request addressed to the new virtual circuit. If the other side supports InARP, it may return a reply indicating the protocol address requested.How can the other side reply with the protocol address requested?Where does it get this intelligence from?I assume they mean with "Frame Relay station" the (DTE) router.Cisco da Gama, can you "translate" this part of RFC 1293 into easy to understand language for a CCNA student?Maybe this isn't possible. ;-)Thanks.

Well, thanks for the link but I'm still not getting it.

I read for example this:

7.2. Protocol Operation Within Frame Relay

One case where Inverse ARP can be used is when a new virtual circuit is signalled. The Frame Relay station may format an InARP request addressed to the new virtual circuit. If the other side supports InARP, it may return a reply indicating the protocol address requested.

How can the other side reply with the protocol address requested? Where does it get this intelligence from? I assume they mean with "Frame Relay station" the (DTE) router.

Cisco da Gama, can you "translate" this part of RFC 1293 into easy to understand language for a CCNA student?

Maybe this isn't possible. ;-)

Thanks.

[edit] Sorry for the text formatting problem, Outlook Express freaked out on me. [/edit]

Let's take a concrete example. Imagine a point-to-point FR link like this

A (DLCI = 102) ------[ FR CLoud ]------- B (DLCI = 201)

Let's say A's IP address is 10.10.10.1 and B's IP address is

10.10.10.2. When the circuit first comes up, B send an inverse-ARP packet with his own IP address out onto the FR link. A sees a packet from 10.10.10.2 come in on his own link with DLCI 102. So, he can set up the following map

IP address = 10.10.10.2, DLCI = 102

Note how he is mapping the remote IP address to his local DLCI. This is sorta the opposite of the usual ARP with ethernet for instance and that is why this is called "inverse" ARP.

Now, when A sees the inverse ARP packet come in, he sends out an inverse packet of his own so that when B sees it he can set up the following map

IP address = 10.10.10.1, DLCI = 201

Now, the layer3 address to layer2 address mapping needed is complete and A and B can communicate with each other. If you are familiar with gratuitous ARP in ethernet, the section you referered to in the RFC is similar to that mechanism for frame relay.

Cisco da Gama

Sorry I've just notice this post. I hope that this example by Cisco da Gama explains InArp ;)

------------------------ Ern

Thanks for clearing this up! Very helpful.

That was a very good explanation of Inverse ARP. However, I still have a couple of lingering questions that I can't seem to get around.

Consider an example as shown below:

A {DLCI=102} -------- FR CLOUD 1 -------- B {DLCI=201} --------- FR CLOUD 2 -------- C {DLCI=301} IP=10.10.10.1 IP=10.10.10.2 IP=10.10.10.3

Initially lets say that A & B are active and C is down. A thus knows that DLCI 102 can be used to reach 10.10.10.2 via FR Cloud 1. B knows that it can reach A through DLCI 201 by traversing FR Cloud 1

C now becomes active. QUESTION 1>> Is it possible to create DLCI 301 such that A can be reached by traversing FR Cloud 2, Router B and FR Cloud 1? (B would be acting like a FR switch in this case).

If it is indeed possible to create a DLCI at C as explained above, will Inverse Arp allow C to know that the new DLCI (Say 301) maps to

10.10.10.1; i.e. will the InArp packet be allowed to traverse B?

I would really appreciate if you could kindly help me out with the same.

Regards,

- PinkFloyd

No, this cannot be done.

Since the answer to Q1 is "no", this does not apply.

You are unnecessarily complicating the situation here. What is happening is that you are trapping yourself at layer2 in thinking about this problem. You need to move up to layer3 now.

For the routers in your network above to communicate, routing needs to be turned on. With routing, C will know that to get to networks behind A it should use B as the next-hop and it already has a FR map to get to B and so things will just work. It will not need a FR map for A or networks behind A etc.

Cisco da Gama

Excellent!!

Thank you Sir for your time and efforts!!

Apologies for the delayed reply!!