>>What does 'subnet mask' mean, anyway?
It's what a subnet wears to a fancy-dress ball.
BUT if you mean an IP subnet mask, that's a little different.
An IP subnet specifies a collection of IP addresses that
are locally connected (on the same subnet), that is, the
addresses that can be reached without going through a
gateway.
Subnet masks are 32 bits and consist of 1's followed by
0's from left to right, that is, it's illegal to
have a 1 after the first zero. The zeros of the mask
specify the variable part of addresses in the subnet;
the ones in the mask specify the fixed part.
For instance, my local lan has a subnet mask of
255.255.255.248, which is 111...111000, and my
assigned IP addresses are xxx.xxx.xxx.240 - 247.
(I'm not telling the xxx's, but they're behind
a firewall anyhow <g>.) So anything I send from
a node on this lan going to xxx.xxx.xxx.240 through
xxx.xxx.xxx.247 go to an address on the subnet as
determined by the mask. It matches because it
is the same as my address in all the bits that
are 1 in the mask and differs at most in bits
that are 0 in the mask. So IP sends it
out on the local lan. If it doesn't match
my machine's address in some bit that's 1 in
the mask, IP sends it to a gateway for routing
through the IP network (the gateway better have
an address on my subnet).
Here's another way to think about it (sorry you
asked yet?): Your IP address is X, your mask is M,
and you generate a packet to go to IP address Y.
IP does a logical AND of the mask with X and compares
that to the logical AND of the mask with Y, that is:
(X AND M) compared to (Y AND M)
If these are equal, X matches Y in the positions that
are 1 in the mask and are on the same subnet. If not,
they are not on the same subnet and the packet must
be routed. |
