I'm currently learning the basics about frame-relay, and everything makes perfect sense so far. There is one thing I just don't get though.
Lets say we have three routers going through a frame-relay switch. R1 is connected to two other routers, R2 and R3. We are using multi-point, split-horizon is turned off on the WAN interface on R1, and EIGRP is running on all routers.
At this point all three router's routing tables are good to go, but if R2 pings R3 it will fail and vice versa. This is, of course, because both routers (R2,R3) have R1's IP address mapped to their local DLCI. They don't have the router's IP that is not directly connected mapped to the DLCI, so when they recieve the packet it is dropped because it doesn't know how to respond to the source IP address,even if it's on the same subnet. The solution is to map that IP address to the appropriate DLCI. Now everything works fine, no problems.
Thats the part that makes sense to me. The part that doesn't is why does mapping that IP address to the DLCI work for ALL networks after that? If there is a second network beyond our first one, it doesn't have to same problem our first network did.
Basically,
- R2 recieves a packet from a router in Network 1. It doesn't have to source IP mapped to a DLCI, so it is dropped.
- We map that IP to the DLCI, and everything is fine.
- R2 recieves a packet from a router on Network 2. It doesn't have the source IP mapped to a DLCI, but it knows to send the response to the appropriate DLCI anyway. Why is that?
Is this an encapsulation issue?
Thanks in advance for the help.
