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

Well, I am not trying to hijack the entire catagory, but I figure going to the experts is needed.

So I need some clarification of the way IP address and Subnets work together in regards to private networks.

So in a private network, The most IP addresses you can have in any one range is 255, correct? Meaning 192.168.0.1 thru 192.168.0.255. All machines in this range can send data back and forth to each other. They all have a subnet of 255.255.255.0

Now lets say you needed more addresses than just the 255. Lets say you had 500 computers or even 500 IP phones that needed to talk to one another. What do you do? How exactly do the IP range and subnet range work together. I have done some online research, but it still isn't clear.

Now the second part of this question is you have 500 computers, but 50 of them are in HR and they still need to communicate, but they need protection too. Is this where they get there own SUBNET ?

I appreciate anyone who can enlighten me.

Z-man

In theory you can increase a class c network past 254 however, the best way would be to change to a class b ip scheme.

I don't really know---for any of my IT questions--I just hand it over to JJ from our sister company (Rockford IT). But from what he has told me when he "baby talks" to me---you change the IPconfig to mask being 255.255.0.0--then you can have a maximum of 64514 computers. The range would be 192.168.x.x to 192.198.254.254.

Now, I don't know if this is correct, because I'm just remembering, here. A couple of years ago, he took pity on me and slowly went through the "math" of IP addresses with me.

Hopefully, someone who really know their stuff can add to this.

This is a very complex subject that I could spend days writing about. Id recommend going to Youtube and searching for IP address basics. There you'll find hundreds of videos covering just about everything about TCP/IP.

For the first question yes you could increase the size of the network by subnetting or switching the network class.

The second question is to protect or isolate certain groups of computers you would use VLANs.

The subnet mask tells the device whether the address its trying to reach is local. It does this by looking at its own IP address and at the subnet mask. If the subnet mask is 255.255.255.0 for example then the device knows that the first three octets are local if they are identical to the devices IP address. If the device has an ip address of 192.168.0.25 then 192.168.0 is the local network. 192.168.0.2 192.168.0.58 192.168.0.201 are all local. If the device is trying to reach 192.15.0.1 then it looks and sees that the second octet is 15 and not 168 and it knows that this address is not local and must be sent to the gateway. If the subnet mask was 192.168.0.0 then now you have the possibility to have 65536 hosts because everything from 192.168.1.1 to 192.168.254.254 is local to a pc having an address of 192.168.x.x.

so, if on my local network, I put in a subnet of 255.255.255.248, does this mean I can only use the first 248 IP addresses?

Also, did I mis-speak when I said a limit of 255 addresses, when it is actually 256 because you count 0 as an address so 0-255?

0 and 255 are not typically used for addresses so you would use 1-254. There are a couple of special cases but for your purposes just know that you can use 1-254. If you use 248 you can only have 8 hosts. That's subnetting and more advanced than basic addressing. To keep things simple for you just know that the 255 means that octet on the device must match. so 255.0.0.0 means that a pc with an address of 192.x.x.x is on the same network. Any device 192. something is on the same network but any other number in the first octet means its a different network. So an address of 191. would be on a different network. For most purposes you would use either an a, b, or c subnet mask which are 255.0.0.0 255.255.0.0 and 255.255.255.0

If a network needs more then 254 hosts (devices), they will change the network to class B network. I have hear that supernetting is possible to get more IP address withing a network, but most will aether use VLans or use a Class B addressing. I have not actually seen a class C used with a class B subnet mask. All the network classes I took (online and in class)mention supernetting or Classless routing, but don't say why or why not to do this.

Anyway, I aggress with tito that instructions for subnetting could go on and on.

In the case of the OP network, supernetting the Class C by changing the subnet to 255.255.0.0, will allow him to use IP addresss 192.168.0.1 through 192.168.255.254 as usuable IP address, allowing 6553 IP address.

Usually the network is changed to a Class B network, using IP address 172.16.X.X, with subnet of 255.255.0.0, IP range of 172.16.0.1 to 172.16.255.254 allowing the same number of IP addresses and following the Subnet rules.

Hmmmmmm. You have two questions really, a technical question, and then a methodology question. I can answer the first one but the second one is all up for wild debate. Might as well be asking favorite colors

First we need to establish some references. An IP is made of a 32-bit integer commonly written as 4 8-bit integers in dotted notation. Each one of these 8-bit integers is referred to as a Class. So given that, an IP is A.B.C.D respectively to their class. The second thing is that everything is a binary representation. The short version is that everything is on a base-2 scale. For this reason, netmasks will always start with 0, 8, 16, etc. You will never have a netmask of 7.

A netmask in common form works as an inverse of your usable IP range. Each class (section) of an IP range consists of an 8-bit integer. The maximum value of this integer is 255 (256 digits, counting 0). What each section of a netmask is doing is providing a range within that class of the IP. The range is determined by subtracting the netmask number from the maximum number (256). For instance, if a part of the netmask is 255, that means the range is 1, or equal to the correlating section of the machine's IP.

So if we have '192' in our IP, and it's correlating section in the netmask is '255', that means the range is only 1 integer value, or '192'. Now if we have '192' in our IP, and it's correlating netmask section is '248', then we take 256-248=8. This gives us the range of our IP. So now we know we will have 8 IP's, and they will be in multiples of 8.

So now that we have a magic number of 8, we need to figure out exactly where our IP range starts and stops at. So you can spend a lot of time doing a math formula or you can just wing it like me. Basically when you wing it you are just doing multiples of 8 in your head until you find two neighboring multiples that include your IP. For instance, 8x24=192, and 8x25=200. Bingo, our IP is in a range of 192-199. The reason it's 199 is because 200 is the start of the NEXT range. If you do have a calculator, you can always divide your IP by your Multiple, and the whole-number part of the answer is the starting point of your range. For instance, 192/8=24, so I know my range is between 24th and 25th multiple of 8. If we pretend my IP was 194 instead, then we would get 194/8=24.25. We still get '24' as our starting point.

The 'data' and 'broadcast' IP's are always the first and last IP on your range. This is usually 0 and 255 with a full-class mask. In addition, the gateway is usually the first usable IP in your range.

So given that, here's some quick examples of those things applied:

Example: 192.168.200.0/255.255.248.0

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.


Example: 67.153.177.49/255.255.255.192

A=67
B=153
C=177
D=0-63 (256-192=64, 49/64=0.7, 64x0=0, 64x1=64)

Our usable range is 67.153.177.0-67.153.177.63


I'll go over CIDR notation as well since it's being used quite a bit lately. All CIDR is is the raw bit form of the netmask. An IP consists of 32-bits. A netmask contains 32-bits. All CIDR is is the number of bits in a netmask that are 1 instead of 0. So the higher the CIDR, the smaller the range. So 255.255.255.248 = "/29" in CIDR form. The raw bits would contains 29 ones, and 3 zeros.

Just remember this is a binary mask and you are actually calculating the inverse of it. I usually just loosely do it in my head, knowing that every multiple of 8 = 255 in the netmask. So a /29 = 8+8+8+5. That means my netmask is going to be 255.255.255.X. To figure out X I take 5-8=3 (3-bits of IP range), so 2^3=8(binary to integer), then 256-8=248 (Remember, we are dealing with an inverse mask). That means my last part of the netmask is 248.

And for going the other way, we take 256-248=8. Then then solve 2^X=8, or 3. Then subtract 3 from 8 and end up with 5. So 248 from a netmask is a value of 5 in CIDR.

You can see more details of CIDR and netmask and how it all relates at wikipedia if you want: https://en.wikipedia.org/wiki/Classless_Inter-Domain_Routing#CIDR_blocks

The more VPN's you have, the more important supernet's or CIDR becomes. Lets say you have two offices, one with 100 computers, and one with 250. Office A is 192.168.10.0/24. Office B is 192.168.20.0/24. Without using CIDR, you could not add those 50 new stations to Office B without deploying routers to route traffic internally. That is assuming the solution is to only go from Class-C to Class-B netmask. I don't like that idea, as routers are just another point of complexity, failure, and bottleneck. For me, the solution is to change Office B from a /24 to a /23. That gives me 192.168.20.0-192.168.21.255. But while I was at it, I would probably just change it to a /22 or /21 and be done. I would then worry about changing subnets again when there were more then 768-computers in a single building network. And with the exception of routers, switches, and servers, you better believe I would be using DHCP so I wouldn't have to touch every machine in the process

But, that's just my opinion on it. Like I said in my other post, we might as well be debating our favorite colors. There is no clear right or wrong method

Thanks James for the very concise explanation. My head hurts. To all of those kids out there who never thought that math matters in school, think again. I can't tell you how many times that I wish I had paid more attention in Geometry class as well.