CIDR, class limitations, the subnet calculator, and me.

Terikan

Ok so the subnet calculator is calculating set possible subnets and hosts based on the class of the first octet I'm putting in.

Does the word 'classless' not mean what I think it means in CIDR?

Is there not a way to say, use 192 octet, with /22, and have it borrowing 6 bits from the third octet for subnets?

192.168.0.0/22 subnet. 64 subnets, 4*256-2 hosts per subnet
192.168.4.0
..8.0
..12.0

Or is that just not possible?

Is that a real world limitation? Or is it a protocol limitation? Or maybe public IPv4 rules/policy?

Thanks, and it's ok if you treat me like a noob on this and use small words.

Edit: Ok so what the calculator is doing is supernetting. How can a device tell if subnetting or supernetting is being used when the mask and notation is the same for both?

CCIE Wanna Be

You may be better served posting this in the CCNP or CCIE section of the forum, from a purely topic-driven perspective, this is not really a CCNA discussion. I am by no means an expert, wouldn't even say a novice, but I will give your question a whirl. I am sure that if I am wrong, I will be corrected (it's how we learn).

I will caveat this by saying this is from what I understand and it is to my knowledge level only....

So, typically supernetting is used to aggregate (summarize) a block of contiguous IP network ranges. As an example, say for instance Company 123 owns blocks 192.22.112.x to 192.22.127.x (all of which are /24 networks), all these address are assigned to various branch offices, but they all terminate at the companies ISP Router. The company using some dynamic routing protocol could have a single entry for every network, however the networks could also be summarize to one address (192.22.112.0 /20), and be advertised by only that one route for all the routes that are included in that range.

To my knowledge, I don't believe you can "assign" a public ip address that would violate the base class of the the given address, these are just used for route summary only (because somewhere you would eventually assign an address in a space you don't own addresses in). However, for the "private addresses" I'm not really sure, but I suspect it would be the same, otherwise why give you ranges in each class, i.e 10.x.x.x, 172.16.x.x and 192.168.1.x or more likely you would start to encroach on the public address space surrounding those networks.

The devices can tell the difference because the summary addresses are listed in the routing table and not assigned to an interface, your interface must have a valid ip address 192.22.112.0 /20 would not be a valid address on a public interface(at least that is my understanding).

Hope this is helpful (i'm kinda curious myself), anyone please feel free to correct my mistake as I'm sure there are a few.

EdTheLad

Classless routing means the address is no longer classified into A,B,C etc by looking at the first octet. Why this is important you might ask? Classful routing interrogated the routing table differently than classless routing. With classful routing the rib was examined and if the classful network i.e major network was present but a route to the exact subnet wasn't, no match would occur even if a default route was present. With classless routing the default route is always used as the last resort.

Why do you need to know the classful boundaries A,B,C today since we use classless routing? mainly due to how protocols are implemented i.e. manual and auto summaries.

The only issue's ive seen with classless addresses is with ripv2, you cannot configure a manual summary with a prefix smaller than the classful mask. I.e. class A network cannot be summarized with a mask less than 8, B less than 16 etc. Hence RIP does not support suppernetting/cidr.

So in short, you are pretty free to configure addresses on a device as you please, e.g. 8.0.0.1 248.0.0.0, which would end up encompassing addresses 8.x.x.x,9.x.x.x,10.x.x.x,11.x.x.x,12.x.x.x,13.x.x.x,14.x.x.x,15.x.x.x . Wouldn't be practical, but you can do it!