RIP Confusion

Two different features."Auto-summary" is used at major network boundaries wheras RIPv2s ability to send an update with a mask it benificial for varying the subnet length within a classful network address.

RIPv2 by default behaves like Ripv1 and will summarize at the major network boundary.
As an example

R1

R3
10.0.0.1/30

10.0.0.2/30 | 151.98.16.1/30

151.98.16.2/30

RIPv2 on R2 by default will perform an auto-summary sending 10.0.0.0/8 towards R3 rather than 10.0.0.0/30, and RIPv2 on R2 will send 151.98.0.0/16 towards R1 rather than 151.98.16.1/30.
If "no auto-summary" was configured on R2, the /30 routes would be sent rather than the major net summaries.

RIPv1 is classful,it doesnt send the subnet mask in advertisements,RIPv2 is classless,it does send the mask.
When an update is received on a RIPv1 interface, if the update matches the major network address configured on the interface it will take the mask that has been configured on the received interface.This can cause nice blackholes in your network as the local router will enter networks into its route table with possible incorrect mask lengths.This is why on RIPv1 you need to use a constant length mask per every major network.
With RIPv2 the mask is sent so the subnet length per major network can vary i.e. VLSM.

kalebksp

The "no auto-summary" command does not apply to RIPv1.

RIPv1 never advertises a subnet mask (because it's a classful protocol). The remote router figures out which network it's in based on the classful network definitions. So, "no auto-summary" wouldn't change the behavior of RIPv1 because it always advertises classful networks.

When RIPv2 is used with auto-summary it basically acts similar to v1 when crossing network boundaries. It still advertises the network mask, but it summarizes it to what ever classful mask is appropriate for the network being advertised. "no auto-summary" disables this behavior.

Hopefully my explanation makes sense.

EDIT: Ya beat me Ed. (The sentence above isn't implying your explanation doesn't makes sense, rather that I was unsure if mine did.)

Hey thank you both very much. Your explanations have really helped me out.

I think part of my confusion is differentiating between the terms classless and classfull. I'm looking at Odom's book and it says Classfull means viewing the address as having a network part that is determined based on the rules about Class A, B and C...

Classless means to view the address as having 2 parts
(1) The part on which routing is based
(2) This host part

So classfull always applies the default A,B and C subnet masks to an address, while classless allows us to specify our subnet mask?

Since RIPv2 sends the SNM with the Network IDs, but still sends advertisements summarized at the classfull boundary, what if we have a situation like this:

10.1.1.0/24 | 192.168.1.0/30 | 192.168.1.4/30 | 10.2.2.0/24

In this case R3 is telling R2 that it knows about 10.0.0.0/8
R1 is also telling R2 that it knows about 10.0.0.0/8
What happens here?

netwicked wrote: »

Hey thank you both very much. Your explanations have really helped me out.

I think part of my confusion is differentiating between the terms classless and classfull. I'm looking at Odom's book and it says Classfull means viewing the address as having a network part that is determined based on the rules about Class A, B and C...

Classless means to view the address as having 2 parts
(1) The part on which routing is based
(2) This host part

So classfull always applies the default A,B and C subnet masks to an address, while classless allows us to specify our subnet mask?

When talking about classful you really have to break it up into two parts.
1)Forwarding,with classful forwarding when a packet arrives on an interface and the routing table is examined, the inital class bits are checked to see if there is a route to the major network.
With classless forwarding,the complete subnet is looked up with no reference to the classful network.You'll have to do a little reading on this in regards default routing.

2)Routing, Classful routing means the mask configured on all router interfaces must be constant per major network.So if you configure 20.1.1.1/24 on an interface, every other router exchanging classful routes must use a /24 mask with a major network of 20.x.x.x.
If you try and configure a /30 or /25 etc this will not be allowed and your route updates will not be propagated.
You could have two different major networks with different masks as it doesnt violate the rule, example 20.0.0.1/24, 16.1.3.1/30, since these have different major networks the mask can be different.That means if using 16.1.3.1/30 all other interfaces using 16.x.x.x must use a /30 mask and using 20.0.0.1/24 all other interfaces using 20.x.x.x must use a /24 mask.

With classless routing since VLSM is supported i.e. variable length subnetting, the mask can vary so the 20.1.1.1/24 could be on one router while 20.x.x.x/25 /26 /27 etc could be on another.

netwicked wrote: »

Since RIPv2 sends the SNM with the Network IDs, but still sends advertisements summarized at the classfull boundary, what if we have a situation like this:

R1
R2
R3
10.1.1.0/24 | 192.168.1.0/30 | 192.168.1.4/30 | 10.2.2.0/24

In this case R3 is telling R2 that it knows about 10.0.0.0/8
R1 is also telling R2 that it knows about 10.0.0.0/8
What happens here?

R2 will have two routes to 10.0.0.0/8,best metric wins or if the metric is the same it will load balance.If this happens your network is screwed as some or all of your traffic will be blackholed.Due to this you have to be careful when using the auto-summary and assigning masks.Its called a discontiguos network.

Ed is really explaining this well, so try and fully digest what he is saying and refer back to your texts to get a clear understanding.

A few weeks back I was asking the same questions you are, and asked myself so much that eventually I sussed out the problem with summarization. You have got there too, and it shows you're understanding things.

netwicked wrote: »

Hey thank you both very much. Your explanations have really helped me out.

I think part of my confusion is differentiating between the terms classless and classfull. I'm looking at Odom's book and it says Classfull means viewing the address as having a network part that is determined based on the rules about Class A, B and C...

Classless means to view the address as having 2 parts
(1) The part on which routing is based
(2) This host part

So classfull always applies the default A,B and C subnet masks to an address, while classless allows us to specify our subnet mask?

Since RIPv2 sends the SNM with the Network IDs, but still sends advertisements summarized at the classfull boundary, what if we have a situation like this:

R1
R2
R3
10.1.1.0/24 | 192.168.1.0/30 | 192.168.1.4/30 | 10.2.2.0/24

In this case R3 is telling R2 that it knows about 10.0.0.0/8
R1 is also telling R2 that it knows about 10.0.0.0/8
What happens here?

Classful routing, such as RIPv1 doesn’t necessarily mean that all advertisement are summarized, which can be confusing. As explained above, RIPv1 can advertise subnetted routes so long as the network boundary is not crossed. So the route advertised, the interface it is sent out of and received on all must have the same network mask in order for the subnet to be installed in the routing table of an adjacent router. As no subnet mask is sent, RIPv1 cannot support VLSM as Ed described.

Example: when Router A wants to send a subnet as an advertisement, it compares that subnet to the network that the sending interface is on. If it is a different major network it is summarized, if it is on the same major network it then checks to see if it has the same subnet mask… if it does… it sends the complete subnet as no boundaries have been crossed. If it doesn’t, then VLSM is in play and the route is not advertised.

So, if a subnet in the routing table is on the same major net and has the same mask as the source interface the route is advertised. Router B then compares the subnet against the interface the advertisement was received on. If it is on the same major network it applies it’s own subnet mask to the route, since Router A would never have even sent the route if was a subnet with a VLSM, Router B can assume the subnet mask that the link between the two routers is appropriate.

As you have pointed out, if there is multiple subnets which are not directly connected, which I believe is often referred to as discontiguous, routers in between will either have the best route to the major network, or multiple routes to the same major network. In your example, R2 will have two equal cost routes to 10.0.0.0/8. Try configuring that and pinging something on either one of the subnets. Then think load balancing.

I believe this level of understanding is more CCNP than CCNA. Enter 'behavior of RIP and IGRP' into Google and look at the first result. The page helped me a lot and determines exactly when subnets are advertised and when they are not.

Hey Ed & Neeko,

Thanks for all of that. I'm reading through this over and over, and it's slowly sinking it...

The Google hit on 'behavior of RIP and IGRP' is really helping too.

When you guys say "major network," are you referring to the first octet?

netwicked wrote: »

When you guys say "major network," are you referring to the first octet?

It depends on which class the address falls under.

Class A 0-127
Class B 128-191
Class C 192-223

151.25.1.2, major net 151.25.0.0
200.3.2.1, major net is 200.3.2.0
31.5.54.2, major net is 31.0.0.0

By major net I mean the classful network.

So the subnets 192.168.10.32 /27 and 192.168.10.64 /27 are both on the 192.168.10.0 major network.

EdTheLad wrote: »

2)Routing, Classful routing means the mask configured on all router interfaces must be constant per major network.So if you configure 20.1.1.1/24 on an interface, every other router exchanging classful routes must use a /24 mask with a major network of 20.x.x.x.

Thanks. I was kind of thinking that, but didn't know for sure.

I am really trying to clarify classless vs. classfull. I am using both Lammle's book and Odom's.

Odom is saying that Classfull obeys the rules of Class A, B, and C.

Which I understand as Class A uses /8 SNM, Class B will use a /16 SNM, and Class C will use a /24 SNM.

This is where I am confused...

I get Ed's explanation and it's the same thing that Lammle says in his book.

But if we go by what Odom is saying how can we use a /24 SNM on a Class A network and still refer to it as Classful when the default Class A SNM is 255.0.0.0? Once we apply the /24 mask to 20.1.1.1, doesn't that make it Classless?

Sorry guys I am really having a hard time with Classless vs. Classfull Addressing. I have read both Lammle's book and Odoms ICND1. Maybe I'm looking way to deep into it, but its really driving me crazy now.

To me, Classfull means strictly:
IP starting with 1-126 and uses only 255.0.0.0 as a SNM. No other SNM can be used.
IP starting with 128-191 and uses only 255.255.0.0 as a SNM. No other SNM can be used.
IP starting with 192-223 and uses only 255.255.255.0 as a SNM. No other SNM can be used.

To me, Classless means strictly:
IP starting with 1-126 and we may use any mask of /8 or larger.
IP starting with 128-191 and we may use any mask /16 or larger.
IP starting with 191-223 and we may use any mask /24 or larger.

You need to differentiate between assigning an address/mask pair to an interface and how a routing protocol uses this information.
A class A network has a major network boundary at 8 bits, if i add additional bits so instead of /8 its /16 i have added 8 bits of subnetting to the major net number.
RIPv1 is a classful routing protocol so it assumes you will subnet consistently thoughout your network for each major network.Ripv1 doesnt send a mask in its updates, when the farend neighbor receives a route, the route address is compared with the local interface address, if the major network of the route matches that of the received interface it will inherit the mask from the receiving interface.
When a route is received on an interface and if it has a different major net number to the received interface, Ripv1 will automatically assign the the major network mask to the route, this is called auto-summary.
This does not happen with classless protocols as they have the mask sent in the updates.

netwicked wrote: »

This is where I am confused...

I get Ed's explanation and it's the same thing that Lammle says in his book.

But if we go by what Odom is saying how can we use a /24 SNM on a Class A network and still refer to it as Classful when the default Class A SNM is 255.0.0.0? Once we apply the /24 mask to 20.1.1.1, doesn't that make it Classless?

What are his words, exactly?

drew2000

This thread makes me feel a little dumb...

Drew

Thanks again for your reply Ed. I thought classful means that when you see an IP Address, summarize based on what class it is a part of.

Example: If you see 172.18.20.56/25 then the classful address means transforming it into 172.18.0.0 255.255.0.0. So, we are basically removing the /25 mask and writing it as a /16 mask based on rules of Class B addresses (the rule being that they use 255.255.0.0 as the SNM).

Another example: If you see 10.5.196.112/30
Classful means turning that into 10.0.0.0, since by default anything starting with a 10 has a 255.0.0.0 SNM.

Neeko wrote: »

What are his words, exactly?

Odoms words?

This is what Odom said:

"Note that the concepts shown in Figure 5-7, with three parts of an IP address (network, subnet, and host), are called classful addressing. The term classful addressing refers to how you can think about IP Addresses-- specifically, that they have three parts. In particular, classful addressing means that you view the address as having a network part that is determined based on the rules about Class A, B, and C addressing-- hence the "classful" in the term."

netwicked wrote: »

Thanks again for your reply Ed. I thought classful means that when you see an IP Address, summarize based on what class it is a part of.

Example: If you see 172.18.20.56/25 then the classful address means transforming it into 172.18.0.0 255.255.0.0. So, we are basically removing the /25 mask and writing it as a /16 mask based on rules of Class B addresses (the rule being that they use 255.255.0.0 as the SNM).

Another example: If you see 10.5.196.112/30
Classful means turning that into 10.0.0.0, since by default anything starting with a 10 has a 255.0.0.0 SNM.

Yes, but you only summarize when the major network changes.If all routers in your network are using the same major net 10.x.x.x. you dont need to summarize.

---R1

R3
10.0.0.1/30 10.0.0.2/30 | 10.0.0.5/30 10.0.0.6/30 | 151.98.1.1/30

R2 will send an update to R1 with 10.0.0.4, R1 will see the route arriving and check the class,Its a class A route, R1 will then compare the first 8 bits of the ip address of the route received with the ip address configured on the interface.Since they are the same major network the route will inherit the mask for the interface.R1 will now see it has a route to network 10.0.0.4/30, remember the mask was never sent....

R3 will send 151.98.1.0 toward R2, R2 will compare the routes ip to its interface ip, route = 151.98.x.x,interface= 10.x.x.x, major network is different, R2 does not now what the mask for 151.98.1.0 is so it defaults the mask to the classful boundary /16, R2 now has a route 151.98.0.0/16.

I'm done, if you cant grasp this go back and read as i cant explain it any better.

No no, I'm actually getting it now the more I read your explanations.

Your explanations actually help me more than the 2 books do.

I'm running some Packet Tracer labs as well and testing it all out.

Thanks for running through this with me a few times. I know it's pretty basic stuff.

I'm grasping it slowly

netwicked wrote: »

Odoms words?

This is what Odom said:

"Note that the concepts shown in Figure 5-7, with three parts of an IP address (network, subnet, and host), are called classful addressing. The term classful addressing refers to how you can think about IP Addresses-- specifically, that they have three parts. In particular, classful addressing means that you view the address as having a network part that is determined based on the rules about Class A, B, and C addressing-- hence the "classful" in the term."

What part of that indicates an original class A address wth a /24 mask is classful? Classful addressing is exactly as you are describing it and how Odom has described it, you've probably just misinterpreted Odom which is causing confusion.

RIPv1 doesn't change the rules of classful addressing, it just manipulates classless network addresses according to network topology, this is where it gets complicated.

The original classful IP addressing scheme however is as Odom says, based on the rules about Class A, B, and C addressing.

LBC90805

If you bought the TWO BOOK Cisco Press kit, then I would suggest looking into the second book and reading the sections on Classful and Classless routing.

Okay everyone.. i spent a few days with this thread, the Cisco URL posted above, Odom's book, and Todd Lammle's book, and it makes sense! Thank you for all your responses. I really appreciate the help!:)

Good stuff. You will encounter this type of confusion again with other topics, I've had it plenty myself but you always get there in the end.

rwwest7

Also keep in mind that RIPv2 by default is Classful. Until you issue the no auto-summarization command, RIPv2 summarizes routes just like RIP. Only after that command is issued then RIPv2 becomes classless.