subnet propagation with classful protocols

JSW77

please correct me if a am wrong...so this is how i understand the distributing of SUBnet info when dealing with CLASSFUL protocols (rip,igrp): A router will send info about directly connected subnets out an interface if that interface is in the network specified by the "network" command (example- this is the config- #router rip (then) (config-router)# network 10.0.0.0. So in this case, directly connected networks of 10.10.0.0 and 10.20.0.0 etc would be advertised out of any interface in the 10.0.0.0 network, correct?)
BUT, if an interface is not on the 10.0.0.0 network, (say 172.16.0.0), then the subnets (10.10.0.0, 10.20.0.0 etc) will NOT be advertised out that interface, only the classful 10.0.0.0 network will be advertised. if i am wrong, will someone please correct me, thank you.

Webmaster

RIPv1 and IGRP do not carry the subnet in routing updates at all.

JSW77

okay, i have figured out the answer (or at least im 90% sure i have, but feel free to prove me wrong), but let me explain what i was getting at so that you all dont assume i was a fool for asking this question when i already have my ccna. the ccna course just doesnt drill this situation into your head the way they should. assume you have a setup like this:

RouterA

---

RouterB

---

RouterC

---

BorderRouter

---

RouterD

Now say that everything to the left of the BorderRouter is network 10.0.0.0 (routers A,B,C), and everything to the right is network 172.10.0.0 or something. There are 3 Subnetworks to the left of the BorderRouter, they are 10.10.0.0, 10.20.0.0, and 10.30.0.0 (respectively, from left to right). these routers will be configured with the router rip command and then the network 10.0.0.0 command, however each of these routers will also see each individual subnet in its routing table (10.10.0.0, 10.20.0.0, 10.30.0.0). the routing tables passed between these routers include the subnet information because each interface configured with a 10.0.0.0 network ip address also includes a subnet mask of 255.255.0.0. when building a routing table, the router checks the subnet mask of any interface mentioned in the "network" command, and uses that subnet mask as the network boundry to use in the routing table. therefore, these subnets (10.10.0.0, 10.20.0.0, 10.30.0.0) will be advertised out of any interface configured in the 10.0.0.0 network. however, when advertised out of any non-10.0.0.0 network, the route is advertised as the classfull 10.0.0.0 network. i think that the reason this is still considered by cisco to not be routing subnet info is that technically when it uses the subnet mask to make a network boundry decision, it is moving the network boundry to the right (network portion=16 bits, host portion=16 bits) so that it is no longer considered a subnet but just a network. if anyone would like to point out where i have made a mistake here, please do, as i am pretty sure this is all correct but it is possible that i have made a mistake somewhere.

Webmaster

All correct except this part: "the routing tables passed between these routers include the subnet information because each interface configured with a 10.0.0.0 network ip address also includes a subnet mask of 255.255.0.0."

There is no field in RIPv1 or IGRP update packets that allows them to carry subnetmask info.

when building a routing table, the router checks the subnet mask of any interface mentioned in the "network" command, and uses that subnet mask as the network boundry to use in the routing table.
Keep in mind, that it checks local interfaces only. But therefore it knows also the subnet mask of a directly connected router's interface. With a classless routing protocol in your example, there is no way for RouterC to know subnet masks at RouterA, it can only assume by looking at it's own interfaces. And that is why it doesn't support Variable Length Subnet Mask...

JSW77

youre right, i shouldnt say that subnet info is passed by the routing tables, but i guess what i mean is non-classfull routing info (what im trying to say is the route 10.10.0.0 is passed even though the classfull boundry is 10.0.0.0). and i understand what you mean about not checking other routers interfaces, but like you said, in a non-VLSM environment, we can make assumptions that the same subnet mask will be applied to all interfaces on the same network.

Webmaster

yes, although RIPv1 and IGRP do not support Variable Length Subnet Masks, they do support subnetting, the mask just has to be the same throughout the entire classful network.