Take the 2-minute tour ×
Server Fault is a question and answer site for professional system and network administrators. It's 100% free, no registration required.

I have the network address and have to subnet things like this:

alt text

How would I subnet this network to accomodate such a huge number of hosts? :D Any guidance for a newbie? I've already read the How Does Subnetting Work post, but they use small number examples, not big like this.

I have to use CIDR and VLSM.

Thanks guys! :D

share|improve this question
Roman universities are early with the homework assignments this year –  Joris Aug 30 '10 at 16:07
@Joris Erm... he's from Bolivia, not Italy –  Izzy Aug 30 '10 at 17:23

6 Answers 6

A /24 will only fit 254 hosts.

Your only option is NAT, which should just work with about 62 hosts per IP address. You can use RFC1918 addresses in the local network, and have the NAT to external addresses done by the router.

share|improve this answer
up vote 1 down vote accepted

A /24 subnet only holds 254 host addresses. If you have 16.000 hosts, you need a larger subnet. If you really only have then you are asking the impossible.

A /18 will hold 16.383 hosts, which should be enough. Or, if you have two distinct subnets of 8.000 hosts each, a couple of /19 subnets would be better.

Assuming the whole /16 network at Universita' di Roma is at your disposal, you might do this:

 subnet            Broadcast       Netmask     #Hosts  32767  16383  8191  8191


share|improve this answer only equals 254 usable hosts. You will likely need to use a RFC1918 private range and do Natting to the range.

share|improve this answer

You can't. At least not with a /24 net, which is only capable to handle 254 hosts.

You would need at least to handle this, or use private networks with NAT.

share|improve this answer

... for a newbie... the best way to describe the way a subnet works... and know how & why to change it... is as follows.

It's called a mask... because it is exactly that. i.e. what part of the address is important.

Computers understand everything as binary. So, I'll need to break it down a bit for ya.

ip address are 4-bytes separated by a dot to keep things simple. i.e. If you write it as binary... it looks like:


The mask, like I said earlier, is what part of the address is important. i.e. means:


Everything important is a 1... everything not... is a 0. More specifically, If the parts masked by 1s is the same, the address is local. Otherwise, it needs to be routed. (routing is a whole other subject... and I'm not going to touch on it)

so... in this example, as long as the address matches the first 3 bytes... its treated as a local address. (not a "loopback" address... but local to the subnet). So anything that starts with 10100000.01010000.00000001. or 160.80.1. is in the local subnet. Doing some simple binary math... you can see that there are 8 bits that can change and not affect the part that is covered by the mask. 8-bits means 256 values (0-255) Potentially there are 256 unique addresses in that block. 7 unmasked bits would mean 128 values... 9 would mean 512.

You can either expand or shrink the subnet... simply by adding more 0s or 1s in the mask. i.e.: = 11111111.11111111.11111110.00000000 = 11111111.11111111.11111100.00000000 = 11111111.11111111.11111000.00000000

Now... as has been previously mentioned by someone else... there is a difference between the number of usable addresses, and the number of total addresses. For each subnet... 2 addresses are reserved. 1 for the network identifier (the very first address) and one for the broadcast address (the very last). So... as previously mentioned... in a network with 256 address... there are only 254 usable addresses. 128 = 126, 512 = 510... etc.

With this simple knowledge, you can easily build your subnet mask to be as big or as small as you want. I do have some words of caution for you as well...

I noticed you are playing with a network... This doesn't fall within any of the "blackhole" networks, which means potentially could interfere with whoever owns that IP block. If you cause them harm, they can take legal action against you. If you don't own that IP block, you shouldn't use it. There are 3 black hole networks designed for private use, that you can cut-up any which way as long as you keep within them.

I hope this crash-course will stick with ya... and start to open your eyes to a much larger set of concepts in networking. Feel free to pester me if you want to go any farther...

share|improve this answer

Your drawing shows two seperate groups of 8000 hosts, but they are connected to a single router on a single interface and only a switch between them. Unless you are doing something like seperate VLANs on the switch and the router is doing something like a "Router on a Stick" design, then these two groups of hosts would not actually be on separate networks at all.

As stated by several others above, you can't IP 16000 separate hosts with 256 addresses. You network drawing doesn't really go into depth as to what sort of requirements these hosts will need, but if you follow standard real world networking, then all of your hosts would not have public IPs on them, they would be addressed privately and you would use a limited amount of public IPs to provide addresses for devices that provide your NAT services.

share|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.