# Understanding IPV4 private address subnetting

I think I have a decent understanding of subnetting however I'm confused with one little portion pertaining to the private address range 172.16/12

I see a lot of places say that this private address range from 172.16.x.x - 172.31.x.x allows for 16 networks and 2^16-2 hosts.

However this is from the point of view of a /16 mask, rather than a /12 mask which is where I'm getting a little confused.

Wouldn't the network be the upper 12 bits, with the remaining 20 bits for hosts? It seems as though the places saying that it can have 16 networks are automatically subnetting by taking an extra 4 bits from the host portion of the private address range. Why are they doing this? What am I missing?

I see a lot of places say that this private address range from 172.16.x.x - 172.31.x.x allows for 16 networks and 2^16-2 hosts.

In the time before 1990 "classful routing" was used.

This means that there was a rule on how the network mask (or prefix length) was calculated from first bits of the network address:

• 10.x.x.x for example was an /8 network,
• 172.x.x.x was a /16 network and
• 192.x.x.x was a /24 network.

If you (still) are using that rule 172.16.x.x/12 is not intended to be used as one single /12 network, but as 16 different /16 networks.

• Does that mean in the modern era we can subnet starting from /12? for instance if we want two subnets in the 172.16 range we can just borrow 1 extra bit from the host portion and use 172.16/13? Or do most people start subnetting from 16? Aug 13, 2018 at 5:39
• @Eladian With modern hard- and software there is no limitation to use /16 for a subnet. I don't know if it is common to still use /16 or not. Aug 13, 2018 at 5:58
• Perfectly fine explanation but it should be noted that network classes were obsoleted 25 years ago by CIDR. You could simply use the /12 as one large subnet (although it might be too large for practical use).
– Zac67
Aug 13, 2018 at 6:30