How are subnet mask attributed to IP?

I am currently studying networks in general. I read a lot of things about subnets, and subnet masks, but there is one thing I can't understand.

How are subnet masks attributed to IP adresses ? For example : How the network administrator tells that an address A will get a 255.255.255.0 mask, and an address B will get a different one ? (in order to create the subnets).

I looked up into my personal router configuration panel, but I did not see anything.

Solution

A subnet mask is selected by the network administrator to divide a network into smaller sub-networks. There is no rule how to "calculate" it from an IP address.

Let's look at an example with a very common subnet mask 255.255.255.0 which in binary is 1111 1111.1111 1111.1111 1111.0000 0000.

A subnet mask will always have a certain number of zeros at the end. Above netmask has 8 zeros at the end, indicating that there will be 2^8 = 256 possible addresses in the subnetwork. Sysadmin might want to have a smaller sub-network, so (s)he can choose a subnet mask 255.255.255.127 which is 1111 1111.1111 1111.1111 1111.1111 0000 and will allow 2^4 = 32 addresses in the sub-net.

Edit: A sysadmin will assign a subnet mask to each IP address based on the administrative requirements. This is done during the design of the network. Let's say that sysadmin knows that in the new branch there are 100 people in sales, and 20 in accounting. Also sysadmin knows that he has 10.10.10.0-255 reserved for that branch addresses. So one way a sysadmin can divide 0-255 range is:

10.10.10.0-127   : Sales
10.10.10.128-158 : Accounting

Then the IP addresses and subnet masks will be:

IP: 10.10.10.0-127   MASK: 255.255.255.128    : Sales
IP: 10.10.10.128-159 MASK: 255.255.255.224    : Beancounters

In each address range 2 addresses will not be used for users: 1st one because it is a network number and the last, because it is a broadcast address.

This all can look like nonsense work for average home user where you can put all your hosts in one subnet, but once you are dealing with big networks this starts making sense.

Router configuration:

To route data between the subnets you need a router. In your average home router you have 2 routed interfaces - one facing the ISP and another facing your home network which is behind a (usually) 4 ports switch. On enterprise/service-provider routers you will have multiple routed interfaces - each with its own IP address and subnet mask. To make matters more complicated - in enterprise routers you can assign one physical interface to have multiple IP Address and subnet mask pairs.