Ref: 13680013
Title: Using IP Subnet Masks
Date: 6/22/90

Copyright 3Com Corporation, 1991.  All rights reserved.

Subnet addressing is used to share a single IP network address.

The RFC 950 standard does not restrict the user of subnet masks
from selecting contiguous bits of the address.  For example, a
network might be assigned the mask:

       11111111 11111111 00111000 01100000
         255   .  255   .   56   .   96

This mask selects the first two octets, three bits from the third
octet, and two bits from the fourth.  Although it is possible to
arrange addresses to devices using this mask, it makes it difficult
to understand routing tables.  So, most sites choose contiguous masks,
and usually use the same mask throughout an entire set of physical nets
that share an Internet address.

Some ranges of IP addresses are not used.  For example, in the subnet
mask "255.255.240.0," the IP addresses for the third field range from
"xxx.xxx.16.xxx" to "xxx.xxx.239.xxx."  Thus, IP addresses from
"xxx.xxx.0.xxx" to "xxx.xxx.15.xxx" and from "xxx.xxx.240.xxx" to
"xxx.xxx.255.xxx" will get the error message "Illegal IP address."

Do not assign addresses of all zeros or all ones to a subnet mask, because
they have special meanings.  A network address of all zeros means "this
network."  An address of all ones means "all hosts."  Thus, an IP address
"xxx.xxx.3.xxx" with the subnet mask "255.255.240.000" will get an error
message, because the address "xxx.xxx.3.xxx" is interpreted as "xxxxxxxx
xxxxxxxx 00000011 xxxxxxxx" in binary.  The first four bits of the third
field in the subnet network field are all zeros, so it should be a reserved
address.
