Notice that the number of zero bits in the third octet will actu- ally dictate the number of Class C networks in the supernet. Each zero bit makes the supernet twice as large. So, a supernet composed of 8 Class C networks would actually have 3 zeroes (8 = 23).

This would seem very limited since it restricts you to using groups that nicely fit into a power of 2 (1, 2, 4, 8, 16...). How- ever, inconveniently-sized supernets can be accommodated because of a simple fact: a netmask with more 1 bits will over- ride a netmask with fewer 1 bits. This allows a smaller supernet to share the address space of a larger supernet. If, for example, you had a supernet of size 6 and a supernet of size 2, you could assign the larger supernet an 8 network address space and assign the smaller supernet the portion of that address space that the larger supernet was not using. Because the smaller supernet’s netmask has more 1 bits, packets whose address was part of its address space would be routed to the smaller supernet even though the address is also part of the address space dictated by the larger supernet’s netmask.

Addressing Schemes B-7

Page 235
Image 235
USRobotics NETServer/16, NETServer/8 manual Addressing Schemes B-7