Subnet Masks (Basic) - Revisited

[Gretchen]

Throughout the last subnet thread I got a lot of mixed information. Here is
what I think I understand.

**Note - All examples are assuming a Class C network

1.) Based on the RFC you can have an odd subnet mask like 255.255.255.199.
1a.) What RFC is this covered in?
2.) However, in practice that type of subnet mask would not work with many
implementations of IP.
3.) Since odd subnet masks do not work with many implementations of IP I
should just stick to subnet masks that have the network bits then the host
bits.

Thanks again
gl

Juergen Marenda wrote:

> You can mix them because the calculation is just bitwise AND NOT and OR.
> You just dont get a contiuous number space for your subnet
> (for example, you do not have 16, 17, 18, 19, 20, 21, 22, 23 as the
> final
> number in you IP Adresses, but 198, 206, 214, 222, 230, 238, 246, 254
> and instead of having ..16 as "ANY" and ..23 as "ALL" (Broadcast),
> ou have ..198 as "ANY" and ..254 as Broadcast.
> This may be confusing but should work.

Huh? I was purely talking in the binary, not the decimal conversion. In
the binary, the network portion of the address must be continuous, and
the same goes for the host. So, if you have 26 bits of subnetting, the
first 26 bits is the network portion of the address, the remaining 6 is
the host portion. What I mean by mixing is that you can't have the first
15 bits as network, then 3 bits of host, then 6 more bits of network,
then 3 more bits of host, and then end with 5 bits of network.

> Historically (ancient RFC) says the subnetmask ...199 is correct,
> but after the introduction of Classless routing, the n*"1" (24-n)*"0"
> scheme is the only usefull one.
> And many implementations of IP do not work correctly with subnetmask
> like the one above.

Yes, I suppose strictly speaking, this would be correct, and I could see
how an ancient RFC might allow that, but it's definitely not the case
anymore (thankfully, I don't even want to think about how insane subnet
calculations would be then!). Today, you always have a continuous range
of numbers in your subnet.

--
Andrew "Frugal" Dacey,
fru...@tildefrugal.net
http://www.tildefrugal.net/

[Andrew Dacey]

Gretchen wrote:
>
> Throughout the last subnet thread I got a lot of mixed information. Here is
> what I think I understand.
>
> **Note - All examples are assuming a Class C network
>
> 1.) Based on the RFC you can have an odd subnet mask like 255.255.255.199.
> 1a.) What RFC is this covered in?
> 2.) However, in practice that type of subnet mask would not work with many
> implementations of IP.
> 3.) Since odd subnet masks do not work with many implementations of IP I
> should just stick to subnet masks that have the network bits then the host
> bits.

I think the general consensus was that 255.255.255.199 was at one time a
valid subnet mask but is no longer. RFCs do get obsoleted and that's
probably the case here.

[Barry Margolin]

In article <wGaC5.900$m06....@news1.news.adelphia.net>,

Gretchen <gret...@adelphia.net> wrote:
>Throughout the last subnet thread I got a lot of mixed information. Here is
>what I think I understand.
>
>**Note - All examples are assuming a Class C network
>
>1.) Based on the RFC you can have an odd subnet mask like 255.255.255.199.
>1a.) What RFC is this covered in?

RFC 950 is the original specification of Internet subnetting. It didn't
place any restrictions on the subnet mask. IP simply masks the destination
address and the local address each with the mask, and if the results are
the same the destination is on the local subnet, otherwise it's on a remote
subnet.

CIDR changed that. Now instead of subnet masks, we use prefix lengths,
which specify how many of the high-order bits of the address specify the
network or subnet. This allows routing entries to have a well-defined
order, so that when a destination falls into the overlap of two routes the
routing algorithm can tell which should be used (the longer prefix is
called "more specific", and takes precedence).

--
Barry Margolin, bar...@genuity.net
Genuity, Burlington, MA
*** DON'T SEND TECHNICAL QUESTIONS DIRECTLY TO ME, post them to newsgroups.
Please DON'T copy followups to me -- I'll assume it wasn't posted to the group.

[Andrew Dadmun]

Go here: http://www.cis.ohio-state.edu/Excite/AT-rfcsquery.html

and search (local) for subnet mask

Regards,
Andrew Dadmun
Senior Network Engineer - e-Builder, Inc. - http://www.e-builder.net/
MS Exchange Server FAQ - http://www.swinc.com/resource/exch_faq.htm

Post in a newsgroup, read the response in a newsgroup.
*~*~* Questions not answered via email *~*~*

"Gretchen" <gret...@adelphia.net> wrote in message
news:wGaC5.900$m06....@news1.news.adelphia.net...