atcomsystems.ca/forum Forums VOIP Phones & IP PBX Networking IP Addressing and Subnets.

I was good at math, and I knew binary since before high school, (Many years ago) but still not great at figuring it out in my head. Good thing there are free calculators online.

Our office uses VLan with 2 192.168.xxx.xxx subnets. It works good 98% of the time, but it is the 2% that it doesn't that gets real annoying. I am definately a fan of making things simpler. I can VLan a network, but I would rather not if a VLan is not nessesary.

CIDR seems pretty convenient. I am wondering why they would drill classful routing into your head then say that none of it matters anymore. Maybe it is like most other tech out there, using an original method of IP addressing wil be 100% compatable with all devices but CIDR may not be?

Well, it is always good to get a grasp on the basics anyway. I still think it's funny when I see Token Ring or bus topology coax ethenet in these clases.

okay, its starting to make some sense, but barely...LOL.

Kumba, thanks for the clarification of counting backwards on the subnets.
So I understand that if there is a zero in the octet, then the entire range is open, if there are numbers in the octet, you would in essence count backwards to know how many IPs are open.

so in your example of using a subnet of 255.255.248.0 you come up with this..

A=192 (256-255=1 IP range)
B=168 (same as above)
C=200-207 (256-248=8, 200/8=25, 25x8=200, 26x8=208)
D=0-255 (any time you see 0 it means all IPs!)

Our usable range is 192.168.200.0-192.168.207.255.

then if the subnet mask is changed to 255.255.236.0, would my usable range now be 192.168.200.0 thru 192.168.215.255?? am I understanding this correctly?

When I had qwest, they gave me a range of public IP address, 8 to be exact. 71.33.253.0-71.33.253.7, With a subnet of 255.255.255.248
Am I to also assume that this was done solely to restrict my access to only the 8 IP addresses? Then they could assign 71.33.253.8-71.33.253.20 to someone else, and that person would have a subnet of 255.255.255.236? Is this correct?

Again, I appreciate this information. I am starting to gain some understanding of what is going on. A person with an IP address of 192.158.0.1 and subnet of 255.255.0.0 would have 256x256 addresses at theri desposal..that is a lot!!

And one more clarification. IP addresses and Subnets are binary numbers that are converted into xxx.xxx.xxx.xxx correct?

Actually 236 isn't representable as a binary value. So therefore the answer is non-existant.

The standard netmask definition is there to make things easy for people to read. It's a lot easier to do X-Y=Z in your head and know where things are at. It's however not really a good representation of what is actually going on under the hood. CIDR is the opposite. It's exactly how the computer understands it, but not the easiest for humans to read. Therein lies the conundrum.

The answer is that the next netmask after 248 is 240. Basically, the difference from netmask to netmask increases or decreases by a factor of 2 through 8 iterations. The reason there are 8 iterations is because each "class" has 8-bits assigned to it. This is referred to as a binary boundary. They are as follows (Remember, these are inverse mask, so subtract them from 256):

NM #-bits Math
---------------------------------
255 0-bits (2^0=1, so we get 255)
254 1-bits (2^1=2, 256-2=254)
252 2-bits (2^2=4, 256-4=252)
248 3-bits (2^3=8, 256-8=248)
240 4-bits (2^4=16, 256-16=240)
224 5-bits (2^5=32, 256-32=224)
192 6-bits (2^6=64, 256-64=192)
128 7-bits (2^7=128, 256-128=128)
0 8-bits (2^8=256, or all IPs in the range)

So, going from 248, which has 8 IP's in that range, to 240, would double your IP range to 16. Likewise, going from 248, to 252, would halve your IP range. A netmask will never be anything EXCEPT these values. After you see them enough you'll just memorize them like I have. Either that or you'll get good at exponentials in your head

And as far as your second question, yes. If CIDR wasn't utilized, Everyone would be on the same network, and you couldn't split up an IP range to multiple customers. Think of it like pie. Mmmmm, pieeeee. The class-based set-up represents a whole pie. The "bakery" has 256 of them. You have 2000 people who want pie. You don't have enough pie!

CIDR is essentially realizing that not everyone needs a whole pie. Mmmmm, pieeeee. So what you are going to do is take a knife and evenly cut the pies into sections so that way everyone can have a piece.

That's essentially all a netmask does. A way to subdivide an IP range so that it can be distinguished from others and separately routed. If your carrier only recognizes ip's in a .248, anything you try to do outside that won't work. Your network is LITERALLY defined as only being 8-IP's wide.


As far as your last question, essentially yes. A netmask is just a human-readable string of bits. For instance, the netmask 255.255.255.248 is actually:

11111111.11111111.1111111.11111000 (periods added for class reference)

In CIDR, you just count the number of one's and that is your /29. In netmask, you convert the bits into an integer by taking the number of one's as the power of 2. In this case the class-D would be 2^5 or 32.


As difficult as some of these concepts seem, they are a pale comparison to IPv6. There it's 8 groups of hexadecimal numbers. You will end up with IP Addresses that look like this:
2001:0db8:85a3:0000:0000:8a2e:0370:7334

Sounds fun huh? At least we will never have to change IP scheme again in anyone's foreseeable lifetime:
IPv4 - 4,300,000,000 addresses
IPv6 - 340,000,000,000,000,000,000,000,000,000,000,000,000 addresses

I forget the actual number, but I want to say that there was enough for their to be something like 1000 IPv6 addresses per every square inch on the face of the earth.

236 is not a valid number for a subnet. A subnet of 255.255.224.0 would give you a range of 192.168.192.1 to 192.168.223.254
There are only 9 different numbers you can use as a subnet number. 255, 254, 248, 240, 224, 192, 128, and 0.

"Restrict Access" is not realy correct. The ISP assign your range of IP address. Another customer could have the SAME subnet as you 255.255.255.248, and have the next range assignend to them XXX.XXX.XXX.9 through XXX.XXX.XXX.15 (8 and 16 would be unusable).
Again, 236 is not a valid subnet mask

The Subnet mask isn't to restrict access to another network, thats the firewalls job. The subnet masks tells your router or PC wheather the IP address you are trying to send information to is on your local network, or if they need to send the data to another router.


Everything with IPs and Subnets are binanary as far as the netork devices are concerned. They are shown in Decimal numbers so humans can read them easier.

The purpose of subnetting is not to restrict access but to more efficiently distribute IP addresses. Instead of assigning a 24 bit subnet where a customer will end up with 254 addresses of which they may only need 1 , they will subnet and assign a smaller group of addresses thereby conserving and better distributing the finite amount of available addresses. They may for example assign a 29 bit subnet such as 255.255.255.248 for 6 hosts.

You guys have provide a wealth of information here, and I am really begining to get a good feel out of this.

I am assuming the binary number representations are the same for all octets? In other words, if you have a subnet of 255.255.248.0, then the next possible usable subnet would be 255.255.240.0 correct?

And so to revist my original question, if I need 600 IP address, then I would need to have a subnet of 255.255.252.0

this would allow a range of 192.168.1.1 through 192.168.4.255. Did I do this math correctly?

Almost! The subnet 255.255.252.0
will allow up to 1022 IP addresses for hosts (network devices). Subnet mask of 255.255.254.0 will only allow 510. The only problem is the 1st usable IP is 192.168.0.1. You have to start the IP range from there. The IP 192.168.1.1 is within the 1st subnet.

The range of usable IP addresses in the 1st subnet with that subnet mask is 192.168.0.1 to 192.168.3.254. If you assigned a PC with 192.168.4.254 (you can't use the last IP address) then it could not talk directly to 192.168.1.1, because the PCs would think that they are on 2 different network subnets. They would need a router between them if they needed to communicate.


BTW, I have had a subnet calculator open for this entire thread. There are online and free ones you can use.

ahh...so the subnet does dictate what your starting IP address is. So you just cant throw in any range of addresses in the last octet. Rather in this case, since I have changed the value of the 3rd octet, I have to start at 192.168.0.1 and go up from there..

If I only needed less than 254 IP addresses, then I could basically us any private IP range that I wanted and it would not care since my subnet would be 255.255.255.0.

I am sure they make a subnet calculator for my iphone. I will have to look up one. I would always use one for sure, but at least now I understand a basic concept.

I think I have one last question. So I understand that only certain subnet numbers can be represented. Like Kumba said.. 255, 254, 252,248, etc... there is no binary value of 236, as I tried to use in my example. If IP addresses are binary, why are they not restricted as well?

As far a calcuators go, it is still important to know the math and how subnetting works. You could use it to check your work.

It sounds like you are taking a test of some type. I just recently took an online Adtran course, ATSA/IN. My other most recent cert, CTP, was getting about 3 years old.

The Adtran course goes through all this networking in an online slideshow type presentation. The second half is all about specific Adtran products. It is a free course and does not take very long. Like most network courses, online or classroom, you will have to practice subnetting.