Class C address Subnet

control

Really simple question, but have read conflicting info.

An example network - 192.105.105.0 /24

If I wanted to create more addresses out of this due to running out of the available 254 there is no problem with using a /23 or smaller mask?

MrXpert

a /23 would give you 510 hosts on that example and upto 128 sub networks so yes no problem but be aware of the valid ranges. for example 192.105.104.1 to 192.105.105.254

control

Cheers,

I take it the network mentioned is not ideal to use internally as the addresses are from the Public range?

Somnipotent

You are correct. Ideally you'd use an RFC 1918 address space for this scenario. 171.16.x.x - 172.31.x.x will work.

Node Man

control wrote: »

Really simple question, but have read conflicting info.

An example network - 192.105.105.0 /24

If I wanted to create more addresses out of this due to running out of the available 254 there is no problem with using a /23 or smaller mask?

I think in a Class C network, the first three octets are used by the network. So if i understand right, you could not borrow bits from the 3rd octect.

Do i have this right?

networker050184

Node Man, you need to look into VLSM.

Somnipotent

Node Man wrote: »

I think in a Class C network, the first three octets are used by the network. So if i understand right, you could not borrow bits from the 3rd octect.

Do i have this right?

Yes, the valid first octet range of a Class B network goes from 128-191. Class C starts at 192. The first 3 octets are reserved for the network portion and the host bits are available for subnetting. In order for him to achieve the desired number of hosts, he will have to either go to a 191.x.x.x address or one of the RFC 1918 private addressing schemes so he will have 16 host bits to subnet.

networker050184

That is not correct Somnipotent. You can borrow from the network bits using VLSM.

Somnipotent

networker050184 wrote: »

That is not correct Somnipotent. You can borrow from the network bits using VLSM.

Apparently, per Keith Barker, this is done in CCIE lab environments. And I have been shut down

networker050184

You can borrow from the network bits. For example a company could be assigned 192.167.0.0/16. There is nothing stopping them from using this whole range or breaking it down in the third octet to further subnets.

Somnipotent

Right, that makes sense if you're given a /16. In a lab, he could theoretically do whatever he wants, but if a customer is given /24, he can't go and steal a network bit from the ISP and make an /23.

networker050184

No, because someone else is probably already using it! ISPs do it all the time though because they are the ones getting the larger assignments. You can't really change anything about your end assignment though as a customer.

Somnipotent

I guess my confusion lies in overall IPv4 subnetting. You define the class by its most significant bit and subnet according the the specified class. With VLSM, obviously you subnet according to what you're given. Hence, why I say you can steal from the hosts but not so much the network, unless of course you have a class B (/16) space with 24 bits subnetted, where stealing the "network" bit would make sense. I need to get my head out of the classful world.

Good solid discussion here.

networker050184

You could also, theoretically, be assigned 192.0.0.0/12 where bits are 'borrowed' in the second octet of a class C address.

Once you learn address classes for the CCNA you can just go ahead and forget them. They really mean nothing in the modern world of networking.

Node Man

networker050184 wrote: »

You could also, theoretically, be assigned 192.0.0.0/12 where bits are 'borrowed' in the second octet of a class C address.

Once you learn address classes for the CCNA you can just go ahead and forget them. They really mean nothing in the modern world of networking.

!!!!!!!!! *fear* ive spent so much time practicing!!!

Somnipotent

Node Man wrote: »

!!!!!!!!! *fear* ive spent so much time practicing!!!

LOL, dude, you're not alone. I do address allocation sometimes for work and even I got confused. networker is right though. I just haven't seen anyone allocated a /12 aside from ISP who break them down all the way to /30s. Plus, I've been mainly using RFC 1918 address scopes for labbing without thought and plan using classful boundaries.

networker050184

And just for clarification sake, it's more often referred to as Classless Inter-Domain Routing or simply CIDR when dealing with larger assignments that break the class boundaries upward. CIDR is still built on VLSM in general.

Somnipotent

networker050184 wrote: »

And just for clarification sake, it's more often referred to as Classless Inter-Domain Routing or simply CIDR when dealing with larger assignments that break the class boundaries upward. CIDR is still built on VLSM in general.

That being said then 192.105.105.1/23 would be part of the 192.105.104.0 supernet.

networker050184

Yep, supernetting is another term used. In the end it all means the same thing. You can use any mask on any address regardless of the legacy class assignment.

control

This turned into a decent thread!

Just to confirm - I can steal network bits and reduce the subnet mask, from say a 24 to a 22 if this is private addressing on an internal network, and the need arises for whatever reason.

But

Cannot do this if the address space has been assigned by an ISP?

networker050184

It all depends on if that space is already used up or not whether it is private or assigned by the ISP. In theory, yes it is possible, but not always available depending on your current addressing scheme.