Michael

I believe that what Sohail was talking about is that we categorize certain routing protocols (especially RIPv1 and IGRP) as classful and other protocols (such as OSPF and EIGRP) as classless. The classful routing protocols adhere to certain principles such as all subnets being the same size, all subnets within a network being contiguous, and automatic summarization at network boundaries. The classless routing protocols support VLSM, support discontiguous subnets, and support manual control of summarization.

And I must take exception with both of you about the concept that classful networks must use the classful (default) mask. Network 10.0.0.0 is a class A network with classful mask of 255.0.0.0 but subnetting 10.0.0.0 with a /26 mask is perfectly ok and can be classful. I can set up a network to use network 10.0.0.0 and 255.255.255.192 as the mask on its subnets (which is the /26 mask) running RIPv1 as its routing protocol and have everything be perfectly classful. I have seen many customer networks that use network 10.0.0.0 but I have not seen a live customer network that used 10.0.0.0 with just the 255.0.0.0 mask.

You can not look at a single subnet and its mask and know whether it is classful or classless. You must have a larger picture of the network to know whether it is a classful or classless network (if some subnets are /26 and some subnets are /30 then it is classless).

HTH

Rick

Hi Rick

The point you make about being able to use 10.0.0.0 with a /26 mask with RIPv1 is fair enough but surely this only works as RIPv1 does not send any subnet information in it's updates. RIPv1 the protocol is treating this subnet as a /8 even though you may have configured a /26 on the interface of your router. Is this not correct?

The reason that I wrote that when using subnet masks other then the classful mask for the particular network number indicates classless routing is because 99.99999999... times out of 100 this will be the case in a live network. The scenario you outline would surely only be found in a lab environment, though I concede your point is correct.

As always, thank you for the insightful input. It is much appreciated.

Best Regards & Best Wishes for 2008 and beyond to you an yours,

Michael

Michael

NO it is not correct that RIPv1 is treating the subnet as /8. RIPv1 does not send subnet information because it believes that it can use the subnet information of the interface that receives the update. The designers of RIPv1 considered it as an efficiency issue - why should we transmit information in the routing update which the receiving router can easily supply on its own? RIPv1 does not need to transmit mask information because it assumes that the mask will be the same on the receiving interface (because in a RIPv1/classful world all subnets within the same network are the same size).

I believe that it becomes difficult in todays environment to discuss classful and classless differences because so few network engineers today have experience in truly classful networks. Another way to express your point is that in todays environment 99.99999999... times out of 100 the live network is a classless network. You may believe that classful networks are only found in lab situations (and increasingly that is true) but in the early stages of my networking career I worked with many customers whose networks used RIPv1 and IGRP and divided network 10.0.0.0 into subnets using network masks other than 255.0.0.0.

HTH

Rick

Hi Rick

Yet again I am indebted to you for putting my misconceptions straight. It is much appreciated.

I don't know why but I have always thought that "under the hood" as RIPv1 was a "Classful" protocol this meant that it always used the classful mask of the network number even if a different mask was in use.Though when you actually think about it, this is a pretty crazy assumption.

My experience in Networking only goes back a little over 3 years, so as you say classless networking is all I have ever known.

Again, thanks a million.

Best Regards & Best wishes for 2008 and beyond,

Michael

Rick,

That was a nice explanation. I am sure it must have helped many. Just thought I would rate this post.

Cheers

Arav