IP addressing and subnetting.

There are over 100 different protocols in the TCP/IP protocol suite. IP (Internet Protocol) is one of the protocols in the TCP/IP protocol suite. The primary function of the IP is addressing and routing. An IP address is 32 bit long.
Logically, we write the IP address like this: 202.221.100.121 Notice that when we look at an IP address, we see 4 parts, separated by dots. This is because IP has a hierarchical structure. Looking at this IP address we can identify that it's a class C address because the first octet is 202 (between 192 and 223).
When the computer sees an IP address, it does not see any dots that separate the octets. The computer sees the same address like this: 11001010110111010110010001111001
The software responsible for interpreting the IP address (the software that's installed when you install TCP/IP protocol on your computer) determines the class of IP address by looking at the first 3 bits (leading bits) of the address.
Class A address will always start with 0Class B address will always start with 10Class C address will always start with 110
IP address carries two pieces of information: The first part of IP address is a network address, and the second part is a HOST address. Just exactly how many bits of the address represents the Network address is determined by the class of address and the subnet mask.
In our example, computer reads the fist 3 bits of the address and right away determines that it's a class C address. For the computer it means that the first 3 octets (first 24 bits) is a network address, and the rest is the Host address.

Sometimes it is necessary to divide a network into smaller logical groups. This is called subnetting.
Advantages of subnetting are: Reduced network congestion, Increase in network performance.
To implement subnetting, we borrow some of the bits from the host address and use them for the subnet address.
In our earlier example, we have IP address 202.221.100.121, or 11001010110111010110010001111001. The first 24 bits is a network address (because it's a class C network), and the last 8 bits are HOST address. Network: 110010101101110101100100 Host: 01111001
Lets suppose we have a whole class C network assigned to us 110010101101110101100100. We can have 254 addresses available for us to use on our network. If we need to divide our network into 5 subnets, we would need to borrow 3 bits from the host address portion to use them to identify subnet address. We will have only 5 bits left to use for host addresses. We can have only 32 combinations with 5 bits, minus 2 because we cannot have addresses with all 1s or all 0s. So now we can have 5 subnets with 30 hosts on each one, totaling 150 hosts. Additionally, every subnet must have one host address reserved for the router interface because we need to use routers to connect our subnets. Now we down to 29 addresses available for our computers on our 5 subnets. 29x5=145
In the process of subnetting we lost many of available addresses (From 254 down to 145)
In order to implement subnetting, we use subnet masks to let our computers know how many bits from IP address will be used for the network address, and how many for the host address.
The default subnet mask for class C address is 255.255.255.0, in binary it looks like this:11111111111111111111111100000000
In order to borrow 3 bits from the host address part, we must create a custom subnet mask like this:11111111111111111111111111100000 or 255.255.255.224
When deciding on a subnet musk, remember that you need 1 network ID (separate subnet) for each WAN connection in addition to each subnet. What it means is that if you have 2 offices separated by a WAN link, you need at least 3 subnets: one for each office and one just for a WAN link. The WAN link itself will probably only need 2 IP addresses (one for WAN router interface on each router). As you can see, you would waste a lot of IP addresses just for the WAN link (an entire range of IP addresses for one subnet).
Also remember that you need an IP address for every computer, network printer, other networking devices plus a router interface on each subnet. So, if you are planning a subnet, and have 28 computers and 2 network printers, you need at least 31 IP addresses for it (remember at least one for the router interface).
Remember that the subnet mask must be the same for entire network being subnetted. Subnet mask is not a part of an IP address, it is not included in the packets' address header and not transmitted over the network, it is something that you manually assign to every computer on the network. (or can be assigned by the DHCP server along with IP address and default gateway)
Computers look at the subnet mask to determine if the destination IP address is on a local or remote network. If it's determined to be on a local subnet, it will be sent to the destination computer, using destination computer's MAC address (physical address). The MAC address will either be found in ARP cache or the ARP protocol will be used to find the destination's MAC address.
If after looking at destination's IP address and comparing it to the subnet mask, it's determined that the destination computer is on a remote subnet, the packets will be sent to the default gateway (router). Each subnet must have at least one default gateway address, and this address must be local to the subnet.
Incorrectly configured subnet masks can cause serious problems. If sending computer decides that the destination IP address is on the local network, when in fact it's on a remote network, the packet will never be sent to router, and will be dropped after several attempts to find the remote IP on the local network.

Just like IP address, subnet mask is 32 bits long. The number of 1s in the subnet mask is called "prefix notation" . An IP address 219.34.29.114 with subnet musk 255.255.255.240 can be written like this: 219.34.29.114/28 because the subnet mask of 255.255.255.240 has 28 1s and 4 0s (looks like this: 11111111111111111111111111110000)

When planning your subnets, always plan for growth. If you need 500 host IDs on each subnet on a class B network, choose a subnet mask of 255.255.252.0, not 255.255.255.254. Because if you will later decide to add 11 more computers on any subnet, you will have to change the subnet mask and will have to reconfigure every workstation and router on your entire network.
A subnet mask of 255.255.255.128 can be used on the class C network in order to maximize the number of available addresses when subnetting a class C network. In order for it to work, all routers on the network must be able to understand this type of mask.
Cisco router can understand subnet 0, but it must be enabled in the router by issuing "ip subnet-zero" command.The problem with this type of mask is that it only uses one bit to identify the subnet address, and it can only be 1 or 0. Older router may not be able to understand this subnet mask and interpret it as illegal subnet mask.