Comments on Subnetting

In IS 450 we spend some time on subnetting, and everyone should be comfortable with the idea of breaking an IPv4 address into 3 portions: network number, subnet number, and host number. Here's an example class B network with 6b set aside for the subnet number (borrowed from RFC 950 and modified): 
   3                   2                   1
 1 0 9 8 7 6 5 4 3 2 1 0 9 8 7 6 5 4 3 2 1 0 9 8 7 6 5 4 3 2 1 0
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|            Network            |  Subnet   |    Host Number    |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
Here we can see 16b of network number, 6b of subnet, and 10b of host number. Note that for the 10b host number, I said 210 IP addresses, rather than 210 hosts. This is because one of those addresses, all ones, is broadcast on that subnet, and another, all zeroes, is used as a subnet identifier. So we would expect at most 210-2 interfaces on such a subnet.

Historical note:

There are old web pages out there, and web pages using old information. In particular RFC 950 (August 1985) says that the number of usable subnets, with n bits set aside in the subnet portion of an IP address, is 2n-2. This wastes a lot of IP addresses, especially on networks with large subnets. By RFC 1878 (December 1995) this waste was recognized, and 2n was used. It is possible, however, that one could encounter an older device which is in compliance with the older standard, but not the newer.

I continue to recommend the Microsoft TCP/IP Fundamentals for Microsoft Windows, Chapter 4 - Subnetting as a gentle introduction to subnetting which appears to be up-to-date and has no mistakes that I've noticed. Also, see the "Subnetting Confusion" page from an MCSE site.

From here one can go

, Jeff Martens