Neil Croft wrote in a message to Mike Bilow:
 MB>      10.0.0.0        -   10.255.255.255  (10/8 prefix)
 MB>      172.16.0.0      -   172.31.255.255  (172.16/12 prefix)
 MB>      192.168.0.0     -   192.168.255.255 (192.168/16 prefix)
 NC> OK Mike, I have to admit a gaping hole in my knowledge of IP
 NC> here. What is the difference between a class A, B and C
 NC> address?
In the old days, a Class A was the term for an 8-bit mask, a Class B was the 
term for a 16-bit mask, and a Class C was the term for a 24-bit mask.
 NC> Also, how exactly does subnetting down work out? I know my
 NC> employer has 10.16 addresses on a 255.255.248.0 subnet mask
 NC> which gives 10.16.72.0=10.16.79 on my site and I know it's
 NC> something to do with bitmasks but I start to lose it a bit
 NC> here which could prove embarrasing if anyone asks.
IP addresses are simply 32-bit numbers.  They are by convention split into 
some number of network bits on the most significant side and some number of 
host bits on the least significant side.  The idea was that any host can be 
reached through the router responsible for its network, although that is no 
longer true.
Each group of eight bits is called an "octet," and each octet is respresented 
as a decimal number in "dotted quad" notation.  The IP address 10.16.72.19 is 
just the binary number 00001010'00010000'01001000'00010011.  It is often 
easier to represent this in hexadecimal: 0x0A104813.  All are equivalent.
If we now define a netmask of 255.0.0.0, which is 0xFF000000, then all we 
mean is that the most significant eight bits are used to denote the network 
and the rest are used to denote a particular host on that network.  The 
network number is, by definition, the bitwise-and of any host on the network 
and the netmask:
   0x0A104813 & 0xFF000000 = 0x0A000000
In dotted quad notation, the network number is just 10.0.0.0.
Also by definition, the broadcast addess is the bitwise-or of the network 
number and the one's-complement of the netmask:
   0x0A000000 | 0x00FFFFFF = 0x0AFFFFFF
In dotted quad notation, this network's broadcast address is 10.255.255.255.
As less bits are assigned to the network part of the address, there is room 
for more hosts in the network.  For example, with a 24-bit mask, there are 
only 256 possible host addresses (including the reserved zero and broadcast 
addresses).  With a 16-bit mask, there are 65536 possible host addresses.  
With an 8-bit mask, there are 16.7 million possible host addresses.
 
-- Mike
--- 
---------------