Microsoft's view on RFC1812

Everlife

Hi guys,

After reading through the subnetting chapter in the Sybex I've run into another problem. The book says Microsoft has the view that you should not use the first and last subnet in a subnet range, hence making your calculation of the number of networks produced by subnetting to (2^n) - 2.

I hadn't remembered seeing that fact when I read through the MS Press book so I did some research on it. Apparently, this rule was originally defined in RFC950, but was later removed with RFC 1812 (referenced here http://www.tcpipguide.com/free/t_IPCustomSubnetMasks-4.htm). I was unable to find anything on Technet about this. Does anyone know off the top of their head if this is still Microsoft's way of thinking?

dynamik

The question will specify if you are allowed to use all 0s and 1s for subnets. However, they might not state this explicitly, so be sure to read the question carefully.

Everlife

Great, thanks Dyn. Would you advise I assume you're not able to use the 0 and 1 subnets unless otherwise stated?

dynamik

I wouldn't make any assumptions. All of my practice material has actually allowed it thus far. The actually term for this is subnet zero, and you can read more about it here:
http://en.wikipedia.org/wiki/Subnet_Zero
http://www.petri.co.il/csc_ip_subnet_zero.htm
http://www.cisco.com/warp/public/105/40.html

nel

im not sure what MS's stance on this but i would like to know when you find out. Im sure wendell odom says in his ccna library to assume that you are using subnet zero unless stated otherwise because it is set to default on a cisco config.

Now i know this isnt a cisco exam but they are the kings of networking so it will be interesting to see MS's stance for the exam.

dynamik

MS Technet wrote:

Note RFCs 950 and 1122 prohibit setting the bits being used for subnetting to all 1s or all 0s (the all-ones and all-zeros subnets). However, RFC 1812 permits this practice.

http://technet.microsoft.com/en-us/library/bb726997.aspx

Scroll down to table 4.3.

The inability to use all 0s or 1s is a limitation of older hardware and software. It's not an issue on newer equipment. The reason I said to not make any assumptions is that MS exams test your reading comprehension as well as your technical knowledge. Within four paragraphs of text, they may include a small note that the router you're using is not RFC 1812 compliant. Since subnetting is one of your favorite things to do, it is understandable that your excitement might cause you to miss that little bit of information. Be sure to read the questions thoroughly.

Everlife

Makes sense Dyn, thanks. You're probably right in your description of the typical Microsoft question. As long as I know what to look for, I feel confident I'll make the right call.

Thanks for you help everyone.

Slowhand

I would go with the suggestion that subnet zero is assumed to be in use, since the MCSE exams you're taking came out in 2003 and 2004, while RFC1812 was written back in 1995. Also, take a look at RFC1878, which outlines what subnets are valid for IPv4, also from 1995. If you take a look at table 2-1, it states that subracting the two bits from subnet zero and the all-ones subnet is considered obsolete.

*Subnet all zeroes and all ones excluded. (Obsolete)
*Host all zeroes and all ones excluded.   (Obsolete)

The way I figure, if excluding the two bits was obsolete eight years before the exams were written, there's a pretty good chance that Microsoft assumes that you can use the two bits. However, I'd recommend reading the questions on subnetting carefully, a right or wrong answer can be the difference between "subnet zero cannot be used" and "subnet zero can be used" buried in the text.