Today almost all communication takes place over Transmission Control Protocol/Internet Protocol (TCP/IP) version 4, in which each computer has a 32-bit address that is commonly viewed as four numbers separated by decimals.

For example, 10.10.10.5 is an IP address for a machine. Although it is possible to manually configure each machine with an IP address (known as static IP configuration because it doesn’t change), doing so is cumbersome in large environments and is hard to track if machines are reinstalled or replaced.

The Windows Server platform provides the DHCP service, which allows clients, at startup, to send a request over the network for an IP address (a DHCP lease request). A DHCP server that has a database of IP addresses that can be given out, and leases that can be offered, services the request.

The duration of this IP address lease is configurable, but is 8 days by default. At the midpoint of the lease duration, the client attempts to contact the DHCP server and renew the lease so that it can keep the IP address.

If you have computers that connect and then are reinstalled or replaced at the end of the lease duration, the IP address is marked as available again—no IP addresses are “lost” and no manual effort is needed to track them.

In addition to an IP address, the DHCP server sends additional information, such as DNS servers, domain information, and time services.

Each machine, be it a server or workstation, now has an IP address. Humans, however, don’t do well with 32-bit addresses. I’m far more likely to remember juliespc than 10.132.43.175.

Every machine has a “friendly” name, it’s known as the machine’s hostname, but you can’t just talk to juliespc. (That’s not quite true: On a local network, machines have the capability to ask, “Is juliespc out there?” but it’s not ideal to flood a network with this broadcast traffic.)

Therefore, you need a lookup list so that you can check which IP address belongs to a specific hostname. A legacy solution still used today is Windows Internet Name Service (WINS), which maps a 15-character NetBIOS name to an IP address. However, this is being retired as quickly as possible.

WINS allows machines, on startup, to tell the WINS server its name and IP address, and other computers could then ask WINS for the IP address for a specific PC. But WINS is limited and designed for NetBIOS, which is neither a future direction for networks nor an Internet standard.

The domain name system (DNS) is used every day. www.microsoft. com is a hostname that has to be resolved to an IP address.

DNS has traditionally been a manually managed database: The DNS administrators added records for hosts and their IP addresses, which were based around zones. For example, one or more DNS servers host a DNS zone for Microsoft.com.

In that zone is a manually created record named www that resolves to an IP address (possibly more than one IP address in the case of a big web site).

However, the role of DNS grew significantly with the introduction of Active Directory, which uses DNS as the locator service. This meant many new records and types of records in DNS for every domain controller, and managing them manually became too troublesome and prone to errors.

Therefore dynamic DNS (DDNS) was introduced. DDNS is an Internet standard and allows each machine to register its hostname-to-IP-address mappings and additional records that are vital to many services. This also works well for clients that get IP addresses from DHCP and register their hostname-to-IP addresses.