This is kind of a stickler.
Let's say you want to subnet a Class C 198.9.200.0 w/ 1 bit of netmask to
make 2 126(128 lest 2) node networks, it's not allowed(as per the RFC and
many peoples IP stacks)! The minimum subnet you can make is 2 bits which
will make 2(NOT 4) networks of 62 numbers.
e.g.
198.9.200.0 w/ a netmask 255.255.255.192 ( 2bit netmask )
yields
Network 1
198.9.200.0-198.9.200.63 - Which becomes unusable since you cannot refer to
it since 198.9.200.0 is reserved(though rarely used) as a means to broadcast
to all the subnets(if you broadcast 0's)!
Network 2
198.9.200.64-198.9.200.127 - Is a usable subnet 198.9.200.64
Network 3
198.9.200.128-128.9.200.191 - Is a usable subnet 198.9.200.128
Network 4
198.9.200.192-129.9.200.255 - Unusable, same as Network 1, if you broadcast 1's.
It's too bad, since very few people require the broadcast to all the
subnet's capability and it makes subnetting Class C's rediculously
inefficient. Many, Many people do it though and I've run into a few that
have been bitten. Most newer systems seem to function fine, but I've run
into a few vestigial machines that can't live on the broadcast networks(for
lack of a better name).
At 08:05 AM 10/12/95 -0700, Howard Berkowitz wrote:
>>
>> Hi Net&Sys Security poeples,
>>
>> Something that everybody is talking about, but not everybody
>> is saying the same thing about subnetting:
>>
>> Yes, everybody agree that we lose the first and last host of
>> each subnet for net.iding and broadcasting.
>>
>> But, some are saying that I can use all subnet; but others are
>> saying that we lose the first and last subnet...
>>
>> Whom truth is true..
>>
>> Thanks..
>>
>> Yannick Gravel
>> System administrator -- yannick .
gravel @
planet-int .
net
>>
>
>This is a FAQ about a confusing subject. The IP RFC states that
>the all ones and all zeroes subnets are illegal. I recommend
>not using them.
>
>They will work in some circumstances, but can cause obscure
>problems later. A pair of simple examples:
>
> 172.31.0.0
> 172.31.0.0
>
>One is subnet zero of network 172.31.0.0, the other is the whole
>network. Which is which?
>
> 172.31.255.255
> 172.31.255.255
>
>One is the broadcast for the whole network; the other is the
>broadcast for subnet 255. Again, which is which?
>
>You may say that's its easy enough to figure this out if you know
>the subnet mask. "Classful" routing protocols such as RIP and
>IGRP, however, don't transmit the mask in routing updates. If
>the receiving host or router doesn't have another way to learn
>the mask, it can't interpret the address and is likely to have
>a problem with the all zeroes and all ones subnets.
>
>Classless routing protocols such as OSPF, EIGRP, and Integrated
>IS-IS do send mask information, so in principle they can
>support the all zeroes and all ones subnets. Problems can occur,
>however, if they have to export such an address into a classful
>routing environment.
>
>Some routers will let you configure these subnets, others will
>never allow it. Cisco lets you do it only if you explicitly
>configure it; it will give you a %bad mask diagnostic if not.
>
>Howard
>
Andrew Foss Tel. 415/494-NETS(6387)
Network Translation Inc. Dir. 415/855-0725
1901 Embarcadero Rd. FAX 415/424-9110
Palo Alto, CA 94303 email afoss @
translation .
com
web www.translation.com
|
|
