I know one can design a network that connects several IP subnets with Frame Relay. I've come across two main configurations.
Full mesh:
Every router is connected with every other. Thus, I believe that what the diagram is saying is that it is enough to define one IP address for each router that names the connection between the router and the Frame Relay network. One must take into account that all IPs correspond to the same subnet (in this case, 199.1.1.0/24
) to accomplish full connection. So only three addresses and one subnet required for the Frame Relay.
Partial mesh:
Here we don't want each subnet to be able to connect to the others directly, so we must use three different subnets for the interconnections (140.1.1.0/24
, 140.1.2.0/24
and 140.1.3.0/24
). So in this case we would need six addresses and three subnets for the Frame Relay.
I've come across this configuration however, which is some kind of a mix between the previous two:
I believe that in this case one should do the following:
- Assign three different subnets for the Frame Relay network. Let's use for example
140.1.1.0/24
,140.1.2.0/24
and140.1.3.0/24
. - The three links in the A-B-C loop should belong to the same network, as they are an example of a full mesh. They could have the addresses
140.1.1.1/24
,140.1.1.2/24
and140.1.1.3/24
. - The other two pair of links (A-D and A-E) should belong to different subnets. We could use
140.1.2.1/24
and140.1.2.2/24
for A-D, and140.1.3.1/24
and140.1.3.2/24
for A-E.
In a nutshell, we would need seven addresses and three subnets for the Frame Relay configuration.
My question: is everything stated in the question (mainly the number of addresses and subnets used in each of the three cases) correct? I'm afraid I might have a wrong understanding about how Frame Relay and IP add up.