<(E-Mail Removed)> wrote in message
news:(E-Mail Removed) oups.com...
> Alex Fraser wrote:
[snip]
>> Anyway, what I wrote was nonsense - what I meant was that only machines
>> with (using your example) 192.255.255.0/24 addresses would be able to
>> access the Internet through the router. I don't think it matters what
>> their subnet mask is.
>
> Well, it mattered.
Sheesh, I really wasn't thinking clearly last night, was I?
Without other routers in the mix, any given host will only be able to
communicate fully (which implies bidirectionally) with hosts in the range
implied by its address and mask. This is fundamental.
In your situation, provided the address of the router's LAN interface (with
the fixed mask of 255.255.255.0) and the address of the hosts needing
Internet access via the router are in the same /24 address range, using a
mask shorter than 255.255.255.0 on those hosts should not affect Internet
access. For example, 255.0.0.0 should work. (This is what I was trying to
say in my previous two posts.)
Using different masks on different hosts, but with all addresses within the
range implied by the host(s) with the shortest mask, is at best not
particularly useful. This is because it may limit which hosts can
communicate fully with each other, depending on what addresses you use.
Suppose you used addresses in 10.0.0.0/8 with mask 255.0.0.0 for some hosts
and addresses in 10.255.255.0/24 with mask 255.255.255.0 for others
(including the router), which if I understand correctly is equivalent to
what Tony originally suggested. Then IP broadcasts should work between any
two hosts. However, as a consequence of the fundamental point above, a host
that has an address outside 10.255.255.0/24 (mask 255.0.0.0) would not be
able to communicate fully with a host that has an address within
10.255.255.0/24 and mask 255.255.255.0. This is because the second host does
not consider the first as being on an attached network.
Alex