70-216 - Calculating subnets debate

Author

Calculating subnets debate

borsky

2003-02-15, 1:39 pm

There is a big debate going on about how to calclate subnets.

The debate is about whether you have to subtract 2 from the possible number of subnets when sub-divide a network using the formula 2^n-2 (where n is the number of bits) or you don't only when calculating the number of hosts.

Well here is what Microsoft says about this in the Network Infarstructure Design manual (70-221):

"The network prefix portion of your subnet mask is based on the IP address class and is fixed at 8 bits. In this example, your subnet number is set at 16 bits. You can determine the maximum number of subnets by raising 2 to the number of bits assigned to the subnet number (2^16 = 65,536).

NOTE:
When your network designates subnet broadcasts as all 1s or all 0s, you must subtract two available subnets from the total number of subnets. Networks based on Open Shortest Path First (OSPF) and RIP v2 routing protocols use all the 1s and all 0s in the subnet number. If the network in the previous example prevented the use of all 1s or all 0s in the subnet number, your subnet design provides a maximum of 65,534 subnets (2^16 – 2 = 65,534).
You can determine the maximum number of hosts per subnet by raising 2 to the number of bits assigned to the host number minus one (28 – 1 = 255). You must subtract one available host address to allow for subnet broadcasts (when the host address is all 1s or 255).

NOTE
When your network designates subnet broadcasts as all 1s or all 0s,you must subtract two available host addresses. If the network in the previous example used all 1s or all 0s to designate subnet broadcasts, your subnet design provides 254 hosts per subnet (28 – 2 = 254). "

Slinky

2003-02-15, 2:00 pm

So whats the debate?

You seem to have answered yourself on when you do and don't subtract two from the number of subnets.

borsky

2003-02-16, 6:44 am

Slinky,

in our previous argument you firmly stated that calculating the number of subnets you have to subtract 2.

The answer is clear enough to understand: you don't have to subtract 2 unless the network prevented the use of all 1's and all 1's. Networks based on Open Shortest Path First (OSPF) and RIP v2 routing protocols use all the 1s and all 0s in the subnet number, ergo when using these protocols - that is what Microsoft operating systems use - you DON'T SUBTRACT!!!!

For further information check RFC1878

http://www.faqs.org/rfcs/rfc1878.html

or Microsoft Knowledge Base

http://support.microsoft.com/defaul...kb;EN-US;164015

I saw sources that are absolutely misleading about this. The question is which sourceS you consider to be authentic.

gcw123

2003-02-17, 4:37 am

quote:
Originally posted by Slinky
So whats the debate? You seem to have answered yourself on when you do and don't subtract two from the number of subnets.

Agree.

borsky

2003-02-17, 5:14 am

My answer was the conclusion of the debate, which was going on here and in other newsgroups too.
I am only sharing the outcome of the debate thinking it might be helpful.

The problem of calculating subnets comes when you get a subnet question on the exam, like this:

"You are designing an IP addressing scheme for your company's network. You will use the private network address of 192.168.0.0/24 to set up 4 subnets with up to 20 hosts on each subnet. Which of the following subnet masks should you use on your network?"

If you assume that the company's network based on Microsoft Windows 2000 routing protocols which is a logical assumption than the answer is 255.255.255.192
If you don't assume that than you calculate it in the 'old way' and the answer is 255.255.255.224

Unfortunatelly Transcender -the source of this question - is not specific about this and uses both ways of calculting subnets and causing contradictons.

Sponsored Links