Frame Relay question about VC and reachable subnets (pic)

rjbarlow

Hi, all, my question could seem a little foolish, but maybe is not so...
Three routers are attached to a Frame Relay network;
Router1 has PVCs toward both Router2 and Router3, but Router2 and Router3 don't have one PVC between them. All routers are in the same subnet.



A routing protocol provides adequate routes on the three routers and inverse-arp provide correct mapping for all routers.

My question is:

Router2 and router3 can reach each other passing through router1?

My doubt arise from ascertainment of the following fact:
If router2 want to send a frame to router3, he know:
DLCI for reach router1;
Router1 can reach directly the link who share with Router3 and has routes for reach all subnets behind that.

Practically Router2 would have to put into its frame the DLCI to reach router1 and a IP address to reach a subnet behind router3.

Seems to me impossible, but just I would want a support.

Thank You.

Netstudent

Try not to thik of it in terms of your lab. Try to remember that your lab is supposed to be simulating equipment that is geographically separated.

Having all the gear so close together sometimes makes it difficult to understand what would happen in the real world.

But to answer your question Yes 2 would have to go through 1 to get to 3. You only have 2 PVC's in your frame-relay provider's network. IF you bought another VC between those 2 endpoints, then they could go directly to each other. The topology usually depends on the amount of data flow. For example if all your resources and application servers were at router1, then there would be no need to pay for a PVC between 2 and 3.

Also try to understand the separation between layer 2 and 3. Even though a packet may have a destination address of the ethernet network behind router1, the encapsulated packet will be framed with the DLCI to get to Router1's serial interface. When a packet arrives at router2 on it's way to ROuter1, the logic will first check the routing table to see which interface to use. A frame-relay mapping is directly correlated with the routing table's next hop address for a route. Only for next-hop purposes and to decide which DLCI to use for an outgoing interface. Try not to get L3 mapping mixed up with destination ip addresses. The destination IP will be down in the datagram, while the L3 mapping is used to decipher which DLCI to use in a frame. This is only used as a one to one mapping of a DLCI and nexthop IP.


I think with only 2 VC's you would want point-to-point links with different subnets. DTE's that are in the same subnet are generally multipoint.

Not trying to overcomplicate your question, but now you see that your question wasn't foolish at all.

dtlokee

The case you have described is a time you would disable inverse-arp and use static maps. In a partially meshed frame network inverse arp can only find the IP address at the other end of the PVC, this means the spokes cannot use inverse arp to find each other's IP address. You would need to add static mappings on the spokes for the other spokes. Since you only have a PVC from R2 to the hub R1 and from R3 to R1 any traffic from R2 to R3 or R3 to R2 must flow through R1. This means your static maps must on R2 and R3 must use the DLCI to R1 to get to each other.

What routing protocol are you using?

rjbarlow

Thanks for replies guys, both Your answers much helped me to think.

This lab did not preview subinterfaces purposely in order to learning basic frame relay addressing. Is real that inverse-ARP works similar to ARP, but it works a little different because it can't send broadcasts to discover ip addresses.

dtlokee wrote:

What routing protocol are you using?

Only theoretical lab dtlokee, because unfortunately this sim do not allow to create partial meshed F.R. networks.