# While VLSM subnetting, will there be a case that addresses are insufficient? [duplicate]

I am studying network design. My question is will there be a moment where you choose IP address for subnetting but turns out finally that it's insufficient(Not in future, just in present).

The question is better worded as "how to choose IP address for subnetting?"

My question would be how would you decide what network id and what host id aka what class of IP address you should choose for subnetting for different amount of hosts? (I know classes are gone, I am just using it for the term host id and network id, and I also know number of hosts are determined by /## term I am just using it for ease to understand).

For eg-: say you have 4 departments including ISP as follows-:

D1=>120 hosts

D2=> 120 hosts

D3=> 120 hosts

ISP=>5 hosts.

So how would you decide which IP address would be enough beforehand? ( English not my first language and I am unable to explain this properly).

Can I choose any IP address? Yes or No? If yes, is any IP address optimal i.e minimum address losses(I am explicitly designing for academia so no future expansion is assumed)?

What's the logic of choosing the optimum minimum loss IP address?

Say I choose 192.168.0.0/24 a class C ip address( Yes classes are obsolete but I am telling this in terms of network id and host id. It gives 8 bit host address so 256 hosts are possible.but there would be subnet and /## would keep on changing on VLSM.)

Say I choose 130.0.0.0/16 a class B ip address(in terms of hid and nid), what would happen?

How do I calculate that this is my needs, and this is the class of IP address or "this is the IP address" that I should choose? Or don't I need to calculate this at all? And subnetting is infinite host address possible?

I learnt VLSM from here

Relevant question

Zac67 says

For a private network, just use any subnet(s) from the RFC 1918 ranges 192.168.0.0/16, 172.16.0.0/12 and 10.0.0.0/8. There's no one-size-fits-all, but for most purposes, /24 subnets are the most reasonable.

• You can't choose IP addresses/prefixes at random if there's only a remote possibility that you'd connect to the Internet one day. Unless you've bought or leased a range you need to stick to the RFC 1918 prefixes.
– Zac67
Jun 23, 2022 at 9:39
• Also, "VLSM" is an ancient term from the classful era, nearly 30 years ago. Today, it's called CIDR.
– Zac67
Jun 23, 2022 at 9:42
• Sorry, your question isn't very clear but the answer you're giving yourself is about subnetting which is a duplicate. Perhaps you can edit your question for more sharpness and I'll be glad to reopen it. You should note that opinion-oriented question or "best practice" are off-topic here, see the help center. You should also note that NE isn't primarily intended for learning (although it can be used that way) but for network professionals to get expert advice.
– Zac67
Jun 23, 2022 at 10:25
• thanks for giving the reason. i know that my question isn't clear but i am not asking here once this confusion came to my mind. i first asked it in 30 forums already(it's been 7+ days since this confusion came to my mind). also i don't ask in forum to learn. teaching isn't to be taken for granted. just like i don't think a professor can build corporate intranet, same way i don't assume any professional knows how to teach so i prefer learning from academics materials/books etc. i have got most of the answers by now by myself though. i found ip add wastage is same in a.b.c.d or w.x.y.z Jun 23, 2022 at 11:45