Like a street address, devices on a network have unique addresses used to route information to them. There are several methods for establishing and using a server address on a network.
The Open Systems Interconnect (OSI) model for network communications establishes a layered approach to communicating between computers on a network. Each layer of this seven-layer model may have its own way of referring to the other computers with which it communicates. This means that any given network server may have many different addresses on file--for use with the different communications layers at work.
Working at layer two of the OSI model, all systems connected to an ethernet network will have a unique ethernet address. This address is expressed as a set of eight hexadecimal values, such as 00-2B-78-5F-03-8A. This address may be used by network switches and bridges to help constrain and route traffic to this particular server.
The Internet Protocol, working at layer three of the OSI model, establishes a unique IP address for every machine on the Internet. IP addresses have three parts: network address, node address and subnet mask. Though the original design was to have this combination be unique in the world, it must at least be unique on your local network.
IP addresses are typically listed as a set of four decimal numbers, each separated by a period. The left-most portion of the address is the network number for the server. Unlike ethernet addresses, IP addresses are expressed in decimals, such as 188.8.131.52.
The right-most portion of an IP address is the "node" address--the address of a given machine on the network.
The subnet mask is compared to the IP address to identify how much of that address should be considered "network" and how much of the address should be considered "node" address. Given the address 184.108.40.206, for instance, a subnet mask of 255.255.0.0 would mean the network number was 198.178, and the node number would be 58.189.
The basic under-pinning of Microsoft's LanManager protocol, the NetBIOS protocol used a friendly name to identify each machine on a given network. Still in use with Microsoft's Vista operating system, each machine on the network must have a unique node name, such as "FTP-Server" or "PaulDev1." These names are used for binding file and print services, as well as being linked to the IP addresses when the systems are also using TCP/IP as their transmission protocol.