On 31 Mar 2002, at 6:17, [log in to unmask] wrote:

> > If you want to try this but can't decide on some addresses, try using
> > this pair: 10.1.1.1 and 10.1.1.2 with a subnet mask of 255.255.255.0
> > for both computers. Then see if the NET VIEW shows anything.
>
> I just noted the above line in an ongoing thread.  The
> subnet mask given is for a Class C address.  The
> 10.x.y.z IP address is Class A and would need the net
> mask of 255.0.0.0.
>
> Good luck!
>
> Steve Sabljak

  Yes -- and no.

  IP addresses consist of a network part and an idividual host part;
in telephone system terms, you could think of these as the area code
and the local number.
  Unlike telephone numbers (and Novell IPX and AppleTalk addresses),
these parts of an IP address are not separate fixed-size fields.
Instead, the 32-bits of IP address are divided into a prefix that
identifies the local network and a suffix that identifies an
individual host, but the boundary between the two can fall at
different places in the address.

  So, how do we know where the boundary is?
  Originally, the network that became the Internet divided the IP
address space into "classes":  addresses starting with octets 1-126
were designated class A, using eight bits of network address (and the
remaining 24 for host), 128-191 as class B (16 and 16 bits), and 192-
223 as class C (24 and 8).  (There were also designations for classes
D (multicast) and E (experimental).)
  In this scheme, everybody can infer the boundary position from the
first octet.  There are still devices out there and protocols that
follow this scheme, although they've shrunk to a tiny minority.
  Steve is correct, in that a 10.x.x.x address falls in the class A
range, and anything that relies on class-based addressing will expect
10.x.x.x to be a single large LAN with up to 16 million hosts on it.

  That illustrates the major failing of the class-based scheme -- it
reserved half the possible addresses for networks with up to 16
million local hosts, and only an eighth for networks of less than 254
hosts.  There are very very few of the former, and a great many of
the latter (and smaller).  Indeed, what big networks there are tend
to be composed of multiple small nets.
  The solution is variously known as VLSM (Variable-Length Subnet
Masking) or CIDR (Classless Inter-Domain Routing).  Basically, it
lets each network place the boundary between the network and host
parts of the address where it wishes, and provides a couple of
different notations for that.
  The most common notation specifies a subnet mask for each network.
The mask is a 32-bit quantity, like an IP address, and is written in
the same dotted-decimal notation.  Where the bits in the mask are 1,
those bits in the address specify the network, and the remaining bits
in the mask, which are 0, indicate the host address portion.

  So each of the original classes can be expressed as a subnet mask;
this is sometimes called the "natural mask" for an address of that
class.
  The natural mask for a class A address is indeed 255.0.0.0.  If we
still lived in a class-full Internet, we would never have to specify
that.
  But in the Internet of today, it is perfectly legal to have an
address in the "class A" range with some other subnet mask, as long
as (a) all hosts on the same network share the same network prefix as
delineated by the mask, and (b) any nearby routers that need to deal
with traffic for this network know what its mask is.

  The natural subnet mask for class C is 255.255.255.0, and this is a
reasonable mask to use for a home LAN.  The 10.x.x.x block of
addresses is reserved for private local use (i.e., not routed out to
the public net without translation).
  The combination of the two is perhaps mildly unorthodox, but
neither dangerous nor illegal, and should work just fine.

Dave Gillett
CCNA, MCSE, A+

        The NOSPIN Group provides a monthly newsletter with great
       tips, information and ideas: NOSPIN-L, The NOSPIN Magazine
           Visit our web site to signup: http://freepctech.com