On ?, 2003-07-10 at 00:23, Shaul Karl wrote:
> On Wed, Jul 09, 2003 at 03:35:49PM +0300, Amir Sela wrote:
> >
> >
> > > ______________________________________________________________________
> > > Network Broadcast Netmask Hosts
> > > 192.168.1.0 192.168.1.63 255.255.255.192 62
> > > 192.168.1.64 192.168.1.127 255.255.255.192 62
> > > 192.168.1.128 192.168.1.255 255.255.255.128 124 (see note)
> > > ______________________________________________________________________
> > >
> > >
> > >
> > > Note: the reason the last network has only 124 usable network
> > > addresses (not 126 as would be expected from the network mask) is that
> > > it is really a 'super net' of two subnetworks. Hosts on the other two
> > > networks will interpret 192.168.1.192 as the network address of the
> > > 'non-existent' subnetwork. Similarly, they will interpret
> > > 192.168.1.191 as the broadcast address of the 'non-existent'
> > > subnetwork.
> > >
> > >
> > > So, if you use 192.168.1.191 or 192 as host addresses on the third
> > > network, then machines on the two smaller networks will not be able to
> > > communicate with them.
> > >
> > >
> > > \begin{interruptRequest}
> > >
> > > How does the 2 smaller networks know that 192.168.1.191 and 192 were
> > > initially a broadcast and network addresses? Would they treat any one of
> > > 192.168.*.19[12] in the same way?
> > >
> > > \end{interruptRequest}
> > >
> > >
>
>
> I start feeling that someone should tell me straight in the face that
> I should do a lot of reading before posting another message after this
> one.
You need to do a lot of reading :)
>
> > First of all, just to make sure we're on solid ground - you need to
> > realize that once you subnet, the "lowest common denominator" is used,
> > that is, as in the example, if you have 8 bits of hosts, and decided to
> > subnet that to 11000000, you now have 4 subnetworks. Essentially what
> > they did in the example was a two stage operation:
> > A) create 4 different subnetworks -> subnetmask(in our case) is 26 bits
> > B) supernet between two of the newly created subnetworks -> subnetmask
> > is 25 bits on two of the subnetworks.
>
>
> Must I begin with 4 subnetworks and then merge 2 of them in order to
> end up with exactly those 3 subnetworks? The answer is probably yes and
> the reason for this is what I fail to understand.
Practically speaking, you don't need to actually do that. You simply
need to give a /25 subnetmask to the 124 upper hosts, and /26
subnetmasks to the 2x62 lower ones. But as I've stated before, the
reason is simply the way TCP-IP works. With such subnetmasks, 192 is not
a host ID. It's a network ID.
192.168.0.192/26 <-> 1100000.10101000.00000000.11000000/26
| || |
\------------\/------------/\-\/-/
Network ID Host ID
So, from the point of view of 192.168.0.15/26, 192 is a network
number, not a complete IP (network number+host ID which is not all zeros
or all ones). This is why it will be unavailable from the two smaller
networks. If the subnetmask is /25, as it is with the 124 upper hosts,
those IP addresses can be used. But since you would obviously want the
hosts on the two smaller networks to be able to communicate with all
computers on the large network, you can't use these addresses.
>
> > I'm stating this because your question was defocused. the 2 smaller
> > networks know that 192.168.1.191 and 192 are "special" addresses,
> > because they still are. Lets assume that our networks are A,B,CD.
> > When the tcp-ip stack in a host in A initializes, it ANDs its own IP
> > address against the subnetmask to determine its home network number.
> > Then, every outgoing packet goes through the same process. if the
> > resulting bit string is different than that host's own result, the
> > packet is known to be out of this host's network, and the ARP request is
> > made to the default gateway, for forwarding.
>
>
> An ARP request? What for? Is it to find the MAC of the default
> gateway? ARP maps the MACs into IPs, doesn't it? An ARP request would
> send the MAC address and expects to get in reply the IP that correspond
> that MAC, isn't it?
>
Wrong. it's the other way around. When you ping x.x.x.x, the computer
knows the IP already. You just typed it, didn't you? What it doesn't
know is the MAC address of the computer with this IP, so proper
one-to-one communication can't be established. So, an ARP request is
sent to request that the computer with the pertaining IP reply its MAC
address.
>
> > What will happen with 191
> > is that you will be requesting the host on subnet A to send a broadcast
> > packet to network C
>
>
> Why would a host on subnet A consider 191 to be a broadcast address?
> Why it wouldn't consider 192.168.*.191 to be all broadcast addresses?
>
See above. 192.168.*.191 is completely irrelvant here. with subnet masks
of /24 and above, it's evident that the 3rd octet is strictly a part of
the network ID, and as such, has nothing to do with broadcast addresses.
>
> > packet to network C(again, by definition, supernetting 2 smaller
> > networks out of 4 is completely transparent to hosts in the 2 small
> > networks).
>
>
> Why the 2 smaller networks have to consider the fact that CD is a
> supernet? Once again, why 192.168.1.191 is considered by them to be a
> broadcast address while any other IP from 192.168.*.191 is fine?
>
They don't. That's what I've said. supernetting is transparent to
networks not participating in the supernet. This is the REASON why those
191 and 192 addresses are bad to use.
>
> > And with 192, you'll be requesting the host on subnet A to
> > send a packet to the network address of network D. Both of these cases
> > are obviously broken.
> > The funny thing is that the so called CD network does not really exist.
>
>
> According to the mini HOWTO it exists in the routing table of the
> router (section 7.1). It also exists as far as hosts on CD are
> concerned. Nothing mentions C and D separately, the router specifically
> mentions the combined CD and you still claim that the separate networks
> exists but not the combined one. I can't understand that.
The way routers "see" the network is different than the way hosts see
it. A route for the supernet is reported to other routers, yes. That's
also one of the advantages of supernetting. smaller routing tables.
>
> > Supernetting is actually a sort of a hack. you simply remove 1 bit on
> > all the hosts on networks C and D, and it "just works"..
>
>
> Except from the fact that for some hosts (the ones on A and B) 2 hosts
> out of it don't exists.
> Let me put it otherwise. Suppose I was sub netting valid Internet
> addresses, say 192.1.1.0 in a similar manner. Would you, with your
> current IP, be able to contact 192.1.1.191 or would it be regarded as a
> broadcast address as far as you are concerned?
I would have no way of telling(As a host). I just try to ping that
address. Again, giving a 32-bit number, without the context of a
subnetmask, when talking about tcp-ip, is ambiguous. If you're not
giving me subnet mask, you're not giving me the ability to discern what
part of that 32 bit number is a network ID, and what part is a host ID.
it's not for naught that the two most basic and required components for
TCP-ip communication are an IP address, and a subnet mask. without the
subnet mask, it's just a number.
You're the administrator of your network, you call the shots. 191 might
be a valid host id, or it might be a broadcast address.
>
> >
> > To summon it up: from the two other networks' point of view, you're
> > trying to send packets to broadcast and network addresses. And they know
> > it because they're supposed to know it. Their subnet masks tells them
> > that.
>
>
> They should use their subnet masks only for themselves, shouldn't
> they? Why are they making assumptions based on their subnet masks with
> regard to a totally strange network? Why they don't do that for all the
> addresses in 192.168.*.*? Or is making those assumptions for the A (1 to
> 126), or B (128 to 191) or C (192 to 254) Internet address classes that
> one is connected to all that sub networking is about?
they're not. see above. i have no way of telling how remote ip networks
are setup. As a host, I use my subnet mask for one purpose:
distinguishing between destinations that are within my reach, and
destinations requiring me to use the services of a router.
Amir.
=================================================================
To unsubscribe, send mail to [EMAIL PROTECTED] with
the word "unsubscribe" in the message body, e.g., run the command
echo unsubscribe | mail [EMAIL PROTECTED]
