Re: Localtalk gateway/bridge

[dempson@actrix.gen.nz (David Empson)]

william strutts <wrstrutts1@nospam.home.com> wrote:

> "Ken Tucker" <ktucker@NOSPAMiit.net> wrote in message
> news:m1in7.172717$oh1.75160071@news2.rdc2.tx.home.com...
> > > > Class B: 172.16.0.0 through 172.31.255.255 (16 networks with 65536
> > > > addresses each, or larger or smaller subnets as appropriate).
> > >
> > > Are you positive about this one. I checked my Mac ethernet adapter
> > > software and it says Class B is 128.0.0.0.
> >
> > This is correct. 172.16-172.31 are private IPs (non-Internet routable) and
> > 128.0.0.0 is NOT private (thus Internet routable - assuming someone owns
> > that address space).
> >
> > > I had read you only 254 per class.  For instance, 10.0.0.1 to 10.0.0.254
> > > was the only possible range. Are you positive of the ranges you quote?
> > > That information was in my Proxy server manual.
> >
> > It's actually 256 hosts, but as he stated, all zeros (.0) and all ones
> > (.255) are not usable. Kinda hard to explain in full detail here,
> especially
> > if you get into subnetting. Just remember to stay away from 0 and 255 on
> the
> > right-most octet and you'll probably be fine.
> 
> What exactly is the subnet mask for?

An IP address identifies a specific station on a specific network on
either the public Internet or a private TCP/IP-based network.

The mask indicates how the IP address is split between the network
address and station address.  The higher order (leftmost) bits are the
network, and the lower order (rightmost) bits are the station.  The one
bits in the mask identify the network part of the address.

TCP/IP routers are generally only interested in the network part of the
address.  A router is connected to multiple networks through one or more
ports, and it needs to know is which port to use in order to access a
specified network.  Once the target network is reached, the final router
can communicate directly with the target station.

> The manual I had states that it should be 255.255.255.0 but it appears
> from David Empsons info that it would depend upon the Class ranges you are
> using.

Even that is a simplification.  The traditional class mechanism caused
the available IP addresses to be used up too quickly: there weren't
enough class C addresses and too much wastage within class A and B.

A more flexible scheme called "Classless Internet Domain Routing" is
used these days.  This relies on explicit masks to allocate an
appropriate number of IP addresses to an organization.  With traditional
class-based routing, the routers would have determined the mask
automatically based on the address class.

The term "subnet" refers to the concept of an organisation being
allocated a block of IP addresses, then subdividing that block into
logical subnetworks for organisational convenience or to improve network
efficiency.  A simple example might be a company which has two floors of
a large building, which has been allocated 256 IP addresses.  They
decide to define two subnets internally, which correspond to each of the
floors.  An internal router is responsible for transferring data between
the two subnets (floors), but the rest of the Internet only needs to
know about the overall network, not each of the subnets.

In this case, the subnet mask reserves additional bits to specify the
network address.

> So if I am getting this then, a subnet mask for Class A would be 255.0.0.0
> to allow you to use all of the possible address ranges. So a 0 in the mask
> would mean that the value in the IP address can range and a 255 would
> mean that it is fixed? Is this correct or am I missing something here?

Those are the simple cases, assuming you are dealing with a multiples of
8 bits for the network and station parts of the IP address.

The subnet mask can subdivide the IP address at any of the 32 bits, but
is always a sequence of consecutive one bits then a sequence of
consecutive zero bits.

For example, 255.255.255.128 would be a subnet mask for a 25-bit network
and 7-bit station.  Written in binary (with spaces inserted for
readability), it would be 11111111 11111111 11111111 10000000.

> > Most SOHO cable/dsl/isdn routers come with DHCP and prefer the 192.168.1.x
> > network, giving you a range of 192.168.1.1-192.168.1.254. Many people use
> > this. I sometimes recommend the 10.1.1.x network since it's easier to type
> > and remember. I say that but I don't use it myself. ;)
> 
> Why does Microsoft Internet Connection Sharing default to
> 192.168.0.1 then if the standard is 192.168.1.1 for a proxy server?

There isn't any particular standard, beyond the "192.168" prefix.  Our
DSL modem at work uses 192.168.23.*, but I'm not sure if this was its
default or something my boss chose.

The Mac application "Virtual PC" also uses some of these address ranges
internally.  If I remember right, the emulated PC thinks it is running
on an Ethernet with an address in the 192.168.2.* range, assigned by a
DHCP server (provided by Virtual PC), but this is unrelated to the Mac's
own IP address.  Virtual PC acts as a router, forwarding packets as
appropriate.

I expect that Microsoft Internet Connection Sharing is using a similar
technique.


Isn't this getting rather off topic for comp.sys.apple2?  :-)