At 2:00 AM -0400 4/16/01, David Chandler wrote:
>So are we all agreed that there is not a problem with using the all 1s/0s
>subnet?
>The real problem seem to be that they continue to teach it.
A continuing problem. While Cisco has engineers that truly know this
stuff (Fred Baker, for example, was editor of 1812), there's a long
history of training and certification perpetuating old myths. I don't
think it's pure laziness -- there's some culture, I think, that we
can't teach anything that might break an existing customer network,
or that the instructors might have to relearn, or some other reason
for what I would call extreme techical conversion.
>
>I think I can plow through RFCs 791, 950 & 1812 within the next couple days.
>May be there is a more compelling reason than it could cause problems with
>hosts built in the mid eighties.
>
>Chuck stated:
>"Wonder if that's part of the reason I didn't make it to day 2 ;->"
>
>That brings up the real question. What does cisco believe is the correct
>way?
Unfortunately, there is no monolithic Cisco view. Still, if I was in
the lab and didn't see any classful protocols, I'd use these subnets.
Rumor has it that one of the first things you do is prepare an
address plan and show it to the proctor. If this is the case, I'd
point to the zeroes and one subnets in the address plan, cite my
references for using it, and, barring objection, go on.
>It appears that passing the CCIE lab is as much about guessing the "correct"
>way to implement something; as it is raw technical knowledge?
>
>Thanks for your input:
>DaveC
>
>
>"Howard C. Berkowitz" wrote:
>
>> That Cisco page is extremely dated information, and actually not
>> quite right -- RFC 791 is, indeed, the primary IPv4 specification,
>> although the IP address format was originally defined in RFC 760.
>> Neither one of these, however, discusses subnetting, which was
>> introduced later in RFC 950. RFC 760 simply assumed a fixed 8-bit
>> network and 24-bit host field, while RFC 791 introduced classes A/B/C.
>>
>> >Wonder if that's part of the reason I didn't make it to day 2 ;->
>> >
>> >I see the point of the article, but I still believe it is more of a
>> >compatibility issue than anything else. Can't get into the RFC server I
>> >normally use to see if RFC 1812 ventures an opinion. CIDR probably
figures
>> >in here somewhere.
>>
>> CIDR actually is in a set of RFCs, about 1518-1520.
>>
>> Without having it in front of me, 1812 specifically says the all
>> zeroes and all ones subnets are legal, but they can be ambiguous in a
>> classful environment. Their use is quite routine in a classless
>> environment, such as an ISP--I frequently use them in addressing
>> plans and have no problems with modern routing. It's been quite a
>> while since I worked in anything with classful addressing.
>>
>> >
>> >I know that throughout my practice for the lab that I have had
situations
>> >exactly as described in the link you provide. I don't recall problems,
but
>> >then the lab is not reality ;->
>> >
>> >Chuck
>> >
>> >-----Original Message-----
>> >From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]] On Behalf Of
>> >David Chandler
>> >Sent: Sunday, April 15, 2001 12:42 PM
>> >To: [EMAIL PROTECTED]
>> >Subject: Re: designing subnets with all ones/zeros.. [7:695]
>> >
>> >Chuck
>> >
>> >Thanks for the proof read :>
>> >
>> >Bellow is the cisco page & part of the doc relating to zero subnets.
>> >To me it reads "don't; because we say so"
>> >
>>
>>http://www.cisco.com/univercd/cc/td/doc/product/software/ios121/121cgcr/ip_c
>> >/ipcprt1/1cdipadr.htm#xtocid105602
>> >
>> >-----------------------
>> >Enabling Use of Subnet Zero
>> >
>> >Subnetting with a subnet address of zero is illegal and strongly
>discouraged
>> >(as
>> >stated in RFC 791) because of the confusion that can arise between a
>network
>> >and
>> >a
>> >subnet that have the same addresses. For example, if network 131.108.0.0
>is
>> >subnetted as 255.255.255.0, subnet zero would be written as
> > >131.108.0.0which is
>> >
>> >identical to the network address.
>> >
>> >You can use the all zeros and all ones subnet (131.108.255.0), even
though
>> >it is
>> >discouraged. Configuring interfaces for the all ones subnet is
explicitly
>> >allowed.
>> >However, if you need the entire subnet space for your IP address, use
the
>> >following command in global configuration mode to enable subnet zero:
>> >---------------------------
>> >
>> >You mentioned that Windows is not rfc1812 compiant and that it allows
>wacky
>> >subnets and disallows some valid subnets. Was that trial & error or has
>> >microsoft documented this? I hate spending an hour looking for a
document
>> >that
>> >is not there...
>> >
>> >Thanks
>> >
>> >DaveC
>> >
>> >
>> >Chuck Larrieu wrote:
>> >
>> >> Comments within:
>> >>
>> >> -----Original Message-----
>> >> From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]] On
Behalf
>Of
>> >> David Chandler
>> >> Sent: Saturday, April 14, 2001 11:25 PM
>> >> To: [EMAIL PROTECTED]
>> >> Subject: designing subnets with all ones/zeros.. [7:695]
>> >>
>> >> I have two questions regarding using the all ones and/or the all
zeros
>> >> subnet.
>> >>
>> >> Recently one of my co-workers started studying for CCNA and while
>> >> reviewing subnets he kept telling me that you could not use the all
>zero
>> >> or all ones subnet.
>> >>
>> >> CL: classically speaking this is true. Early implementations, etc.
>these
>> >> days this is no longer the case
>> >>
>> >> The Win95, NT, and LINUX hosts didn't have a
>> >> problem with it nor did the routers.
>> >>
>> >> CL: a long time ago on this list we had a discussion of wacky subnet
>> >masks.
>> >> In the course of researching this, I found that the windows IP stack
>was
>> >not
>> >> rfc 1812 compliant in that it allowed discontiguous / wacky / non
>> >contiguous
>> >> ones subnet masks, and that windows also categorically denied use of
>> >certain
>> >> legitimate ip addresses. Such as 172.16.1.255/16 I believe that this
>is
>> > > corrected in Win2K
>> >>
>> >> I tested it with RIP & EIGRP.
>> >> (skipped OSPF since it is classful).
>> >>
>> >> CL: I believe you meant to say "classless" ;->
>> >>
>> >> I found that Cisco and others vendors agree that it will work, but
they
>> >> "Strongly discourage using the all ones or all zeros subnets"
>> >>
>> >> CL: where did you find language about "strongly discourage"?
>> >>
>> >> PS: if some of you try testing this; note that prior to 12.1 you'll
>need
>> >> to enter
>> >> (config)# ip zero-subnet
>> >> before the router will accept a zero subnet on a interface. Starting
in
>> >> 12.1 the zero subnet is enabled by default.
>> >>
>> >> CL: ip subnet-zero
>> >>
>> >> Question #1: What type problems could you run into by using a all
>> >> ones/zero subnet.
>> >>
>> >> CL: issues with older equipment / obsolete equipment / old OS
versions
>> >>
>> >> Question #2: For you folks that are in design; Do you follow or
>> >> ignore the "DO NOT USE ALL ONES/ZEROS" rule?
>> >>
>> >> CL: use both all the time. Of course I sell new Cisco equipment, so
>there
>> >is
>> >> no issue with most customers. Or I sell EIGRP or OSPF designs. Same
>> thing.
>> >> ;->
>> >>
>> >> I'm trying to get a real world idea of what the standard practice is.
>> >> I work at a large corp, so I haven't a clue what sane people do.
>> >>
> > >> CL: so do I and neither do I.
Message Posted at:
http://www.groupstudy.com/form/read.php?f=7&i=765&t=695
--------------------------------------------------
FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html
Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
