Hi every body.
i have few question about following scenario.
Let say we have one central side Rc, and two sides, R1 and R2.
Two sides are connected to central side by vc.
Vc between Rc and R1 uses dlci 1
Vc between Rc and R2 uses dlci 2
Vc between R1 and Rc uses dlci 1
vc between R2 and Rc uses dlci 1
There is no Vc running between R2 and R1.
All traffic between the two sides passes through Rc.
The central side and two sides are connected to frame relay network as shown below:
Rc s0 22.214.171.124/24 -FR
R1 s0 126.96.36.199/24--FR
R2 s0 188.8.131.52/24--FR
routing table at R1:
C 184.108.40.206/24 s0
Let say R1 recieves a packet with destination 220.127.116.11( which is ip address on so of R2).
R1 checks the routing table and find a match out of s0.
Inverse arp at R1 only shows one entry as there is only one vc running on R1.
18.104.22.168 dlci 1
So how does R1 determine it has to send this packet with destination 22.214.171.124 to RC ?
Thanks a lot.
I admit that I didn't explained clearly.
However,there is no need to run a routing protocol you could use static routes for the client subnets on all three routers with the same results.
Here the question is that in FR NBMA a device is able to send a packet to a destination in another remote router client IP subnet, because it maps it to the known DLCI the one to the hub router.
The scenario changes when trying to reach the wan ip address of the other remote router because no resolution mechanism without manual mapping will tell to the router what DLCI to use to reach that ip address. (if no direct DLCI between them so inverse-ARP cannot help)
In second case the ip address is in the same subnet as the serial interface of the router so it is only a question of L2 resolution and there is no real IP routing involved.
So it is possible to have ip connectivity lan to lan and at the same time no connectivity in the FR ip subnet in an hub and spoke topology.
This is the point I tried to put in evidence.
Hope to help
You can actually run routing protocols on the routers over frame relay, and that will make the ping work as the subnet of the host PC on R2 network, will be then knowned by R1.
2. yes you dont have too.
In production networks, I have seen manual mapping used much more than inverse-arp hence thought I will mention it.
I agree that in real world subinterfaces are the better solution
hi why don't you the sub interface. ..
if you are using point to point sub interface just give your local DLCI number.
if you are using the multi point sub interface map your remote ip to your local DLCI number.
in a scenario like the one you are describing without adding a manual mapping as explained by Lucien the router R1 should complain of encapsulation errors that it is not able to find the DLCI to use to reach the ip address.
That is no ip connectivity is in place inside the same ip subnet: this is the core of NBMA issues.
To be noted that remote client lan on R2 can be reached from client vlan on R1 that is what is needed in real world (on the router by using an extended ping with souorce lan ip address) without adding the map statement.
Being able to ping over the FR NBMA subnet from router to router can be useful for troubleshooting purposes.
Be also aware that FR DLCI have range for user traffic DLCI values and that first usable DLCI should be 16. the upper limit depends on lmi type.
values 0-15 are reserved for signalling purposes.
Hope to help