max usable IP address

jackwang

We have the network topology shown in below. Assume RIP v1 is the only routing protocol in use. What is the maximum number of usable IP address that can be supported on each LAN if only one Class C block can be used?

Lan1             Lan2              Lan3
  |              |                 |
Router1------Router2-------------Router3
  |                                 |
Router4------Router5-------------Router6
  |             |                   |
Lan4            Lan5              Lan6

[/code]

the_return_of_the_Ring

I guessed the number of max ip addresses is 96. Correct me if I'm wrong.

You have 12 subnets here and RIP is the only required routing protocol. So subnet mask should be 28 bit, which implys 14 hosts in each subnet. You have 6 lans, added it up to 84 hosts(ip addresses). Plus 6 point to point connections among routers, which make the IP addresses number to 12. 84+12=96. You're gonna waste 2 subnets and, because of RIP you lost those IPs in 6 point to point networks.

I came up with another question. If I want to assign IP addresses to those switchs in Lan1~Lan6, do i need to restrict the addresses to their network address range?

the_return_of_the_Ring

Your question is about the max # in each network. so the answear is 14.

sprkymrk

madonion wrote:

I came up with another question. If I want to assign IP addresses to those switchs in Lan1~Lan6, do i need to restrict the addresses to their network address range?

The short answer is "yes". However, I typically assign an Out of Band (OOB) address to help restrict access. You could potentially use an additional network (you had 2 left) if you configure the routers with additional IP's thus creating an administrative network.

lwwarner

jackwang wrote:

We have the network topology shown in below. Assume RIP v1 is the only routing protocol in use. What is the maximum number of usable IP address that can be supported on each LAN if only one Class C block can be used?

The short answer here is, it depends. It depends on:

1. Do you really want to use RIPv1 to propagate your routes or are you willing to use static routes? 2. Do you have to use that single "Class C block" for the entire network or just the 6 lans? 3. If the answer to 2 is just the 6 lans, then is your block really a class C or is it a /24 from non-class C space? (eg. 172.16.1.0/24) 4. Is your block private or public space? 5. Do you want the answer limited to CCNA-level concepts, or do you want the real answer?

If you have a real class C that you have to use for the entire network, and you really want to use RIPv1 to propagate your routes, then madonion has given the CCNA-level answer above. There are, however, ways to beat that...

Humper

Did you get this question from ****? I had one exactly the same today.

jackwang

Suppose we assigned each LAN a block size of 16 network like this:

LAN1: 192.168.1.0 /28
LAN2: 192.168.1.16 /28
LAN3: 192.168.1.32 /28
......

If we have to use RIP v1 as the only routing protocol, and we know that RIP v1 is classful and does not support VLSM. If we do

Router(config)# router rip
Router(config-router)# network 192.168.1.0

on each router, how do those routers advertise a size of 16 network like 192.168.1.16/28 ? Will they all be comfused by 192.168.1.0/24 ?

pannupandit

well RIP v1 doesnot support VLSM. so what I feel ,routers wont be able to send the packets to the correct subnets and will randomly send packest.but am not sure abt it. if anybody is having any idea do post it.
thanx

EdTheLad

This will work perfectly,when you specify network 192.168.1.0 rip will advertise all subnets belonging to this major network.VLSM means when you have variable subnet lengths in your network as in this example you could have 192.168.1.0/29,192.168.1.8/29 and 192.168.1.16/28 the problem occurs when you use a /29 and a /28 with the same major network number on the same router.Rip will not send updates between these interfaces.When using RIPv1 all subnet masks must be the same length because rip doesnt send the mask it will assume the received update will have the same mask as the receiving interface.

pannupandit

THanx Mr. ed. I just got the thing i wanna know .
gud job

jackwang

Thanks ed, your answer made me much clearer.
Please verify this is correct:

Ripv1 DOES support Classless Inter-Domain Routing (CIDR) but always at a "fixed" subnet mask size, not VLSM. For example if it is 28bit submask, all subnets need to be 28bit masked.

If above is correct, then why Ripv1 is defined as a classful protocol?

Probably is it not related, in what cases we use "no auto-summary" command? Please kindly advise.

lwwarner

jackwang wrote:

If above is correct, then why Ripv1 is defined as a classful protocol?

Probably is it not related, in what cases we use "no auto-summary" command? Please kindly advise.

RIPv1, like IGRP, is considered classful becuase RIPv1 updates do not include a subnet mask. So, when a router receives a RIPv1 update it tries to be smart about it and figure out the correct mask based on local info. Specifically if a network advertised in the update is on the same major net as the receiving interface RIP assumes the same subnet mask as the interface for the advertised route. However, if that little trick doesn't work, then RIP has only 3 choices:

1. Assume a classful mask 2. Assume a host mask 3. Ignore the update

To learn about how and why RIP makes these choices, check out these articles:

Behavior of RIP and IGRP When Sending and Receiving Updates
Why Doesn't RIP or IGRP Support Discontiguous Networks?
Why Don't RIPv1 and IGRP Support Variable-Length Subnet Mask?

The no auto-summary command is mostly useful with RIPv2 in a topology with discontiguous networks. Because of the inherently classful nature of RIPv1 no auto-summary probably won't do what you want if used with RIPv1. Again, see the "Discontiguous Netowrks" article above for more detail.

cj_cheema

Well if some one wanna to implement VLSM /CIDR i think it is better to use RIPv2 rather then RIPv1

CJ

EdTheLad

jackwang wrote:

Thanks ed, your answer made me much clearer.
Please verify this is correct:

Ripv1 DOES support Classless Inter-Domain Routing (CIDR) but always at a "fixed" subnet mask size, not VLSM. For example if it is 28bit submask, all subnets need to be 28bit masked.

If above is correct, then why Ripv1 is defined as a classful protocol?

Probably is it not related, in what cases we use "no auto-summary" command? Please kindly advise.

RIPv1 does not support CIDR,CIDR is supernetting, this is when you subnet the bits to the left of the major network.Here is an example.
192.168.1.0/24 is our major network, its possible for me to subnet this into
/29 chucks and run ripv1 as long as all my subnets have the same length.Now if i wanted to use CIDR i.e. supernetting, that is i want to create a super subnet.I could summarize my routes with 192.168.0/16, the /16 is smaller than the natural /24 class C address.This is CIDR and it gives the possibility to advertise multiple class c major network addresses in a single advertisement.

Ripv1 and IGRP will auto-summarize.This cannot be switched off.The no auto-summary will effect Ripv2 but not Ripv1.