[NTLUG:Discuss] IP super/sub-netting maddness

[Michael Barnes]

Gentlemen, if I may step in here for a moment.  Before blood pressures 
raise to unmanageable levels or people start throwing things or 
contemplating drive-by shootings or something, please let us step back a 
moment and look at the original issue.

Richard has a network he put together with a bit of an error in 
foresight when some configurations were originally set up.  As his 
network grew, those errors became apparent and turned into a bit of a 
snare.  Now he is looking for a viable solution.

With computing and networking, there are many ways of doing almost 
anything.  Almost anything is possible, most things are permissible, 
however, not all of it is beneficial.

Regardless of the RFCs and other fine documents which have been 
referenced, there are several points which must be addressed.  First, I 
think we all agree the existing configuration is indeed non-standard and 
non-conforming to accepted practices.  Yes, it might work, but simply 
working is not always good enough. Second, the standards are applied to 
many pieces of hardware and software, which will be expecting certain 
configurations IAW acceptable practices.  Third, non-conforming networks 
may result in administration nightmares further down the line.

It is always (well, almost always) easier to fix problems now and move 
into the standards conformance, than to haywire it and let it grow into 
the 900 pound gorilla on the coffee table and try to tame it later.

I would like to suggest to Richard that yes, it may likely work the way 
you are thinking of doing it, however, you will continue to run into 
issues down the road.  Further, a question is whose network is this?  If 
it is your home network that only you will deal with, that is one thing. 
  If this is another network that may someday have someone else enter 
the picture to maintain or administer, it would be much better if the 
standards were followed, or those who come behind you may ultimately 
wish physical harm to your person.

To summarize, Richard, what you are proposing may work.  You will almost 
certainly run into further problems in the future.  You will likely run 
into hardware and software problems.  I suggest you seriously consider 
renumbering your network to more standardized subnet groupings.  This 
will save you much stress, hair, irritation and problems down the road.

Okay, lets go get pizza.  Who's buying?

Michael



Wayne Walker told me on 9/20/2006 14:24:
> On Wed, Sep 20, 2006 at 12:01:31PM -0700, joseph beasley wrote:
>> Yes it is, but 255.255.255.200 is not.  128,192,224,240,248,252 are
>> valid. 
> apparently you didn't read the message.
> 
> "http://tools.ietf.org/html/rfc950 shows a mask of 255.255.255.88"
> 
> RFC 950, 951, 791, etc ARE the standard, not books, web pages and
> tutorials written by others.
> 
> netmasks do NOT have to be contiguous high bits to be valid.
> 
> much of the equipment on the net can't handle non-contiguous netmasks,
> but they ARE valid.
> 
> Wayne
> 
>> --- Wayne Walker <wwalker at bybent.com> wrote:
>>
>>> On Wed, Sep 20, 2006 at 11:04:15AM -0700, joseph beasley wrote:
>>>> Don't mean to be pushy....   but here goes...
>>>>
>>>> Odd and invalid.  Here are a few links.
>>>>
>>>> http://www.freesoft.org/CIE/Course/Subnet/6.htm
>>> says 255.255.255.128 is valid
>>>
>> http://www.pku.edu.cn/academic/research/computer-center/tc/html/TC0306.html
>>> says 255.255.255.128 is never valid.
>>> says you can't have 25 bit or 32 bit networks.  You can.  32
>>> bit networks are often used by ip aliases (in BSD I believe) and by
>>> VPN
>>> and Point to Point connections.  I have two racks at ISPs where I'm
>>> allocated a 25 bit subnet.
>>>
>>> Poor choice for verification of anything.
>>>
>>> Neither of them is correct across the board.  Kenneth is right.  Odd,
>>> recommended for Non-use, yes.  Invalid, NO.  Usable in most OS's IP
>>> stacks, NO.  Most IP stacks will consider such a subnet mask as
>>> invalid,
>>> but that is because that is almost never how a network is defined.
>>>
>>>> http://www.exabyte.net/lambert/subnet/subnet_masking_summary.htm
>>> "Are any subnet masking limitations due to Microsoft's O/S's?  No,
>>> it's
>>> inherent in the architecture of DNS."
>>>
>>> subnet masks and DNS have NOTHING in common.  again a site that
>>> someone
>>> threw together to be helpful, but he doesn't know what he's talking
>>> about.
>>>
>> http://freespace.virgin.net/glynn.etherington/subnet_masks_and_ip_for_beginners.htm
>>> "Beginning IP for New Users"
>>>
>>> true.  far from defining the limits of what can and can't be used in
>>> a
>>> netmask.
>>>
>>> NOW, if you can find something supporting either side of the
>>> discussion
>>> HERE then it's meaningful:
>>>
>>> http://www.ietf.org/rfc/rfc0791.txt
>>>
>>> http://tools.ietf.org/html/rfc950 shows a mask of 255.255.255.88
>>>
>>> Wayne
>>>
>>>> --- Kenneth Loafman <kenneth at loafman.com> wrote:
>>>>
>>>>> Just a nit, but it is odd, not invalid...
>>>>>
>>>>> 255.255.255.200 == FFFFFFC8 or 1..11001000
>>>>>
>>>>> which means you have 5 bits to play with, just not an adjacent 5
>>>>> bits,
>>>>> thus there are 32 possible IPs in the subnet, C8-CF, D8-DF,
>>> E8-EF,
>>>>> and
>>>>> F8-FF.  This fits the def of a submask, but would not be
>>> compatible
>>>>> with
>>>>> CIDR notation except as 4 distinct small subnets of 8 each.
>>>>>
>>>>> I've seen this used for device subnets where redundant devices
>>> are
>>>>> subnetted.  Not spiffy, but valid.
>>>>>
>>>>> ...Ken