As presented in previous posts, there are four tools that are used 24/7 in almost every TCP/IP network in the world to help the network layer – Layer 3 – with its task of routing packets end-to-end through an internetwork. These automated tools are identified as:
- Address Resolution Protocol (ARP)
- Domain Name Service (DNS)
- Dynamic Host Configuration Protocol (DHCP)
- Ping
And, every device that uses TCP/IP, in fact every interface on every device, needs a valid IP address. For some of these devices, the IP address can, and should, be statically assigned through configuration of the individual device. Of course, the method used to assign IP addresses to interfaces is, many times, a matter of organizational policies and procedures.
As an example, all of the commonly-used computer operating systems that support TCP/IP enable the user to statically configure the IP address on each interface. For instance, routers and switches typically use statically configured IP addresses. Also, servers typically use statically configured IP addresses. Using a statically configured and seldom-changed IP address helps because all references to that server can stay the same over time.
It’s the same concept as knowing the address of your favorite electronics store. The location never changes and you always know how to get there from your home, on the way home from work, or from somewhere else. In a similar manner, if servers have a static, unchanging IP address, the users of that server know how to reach the server from anywhere, consistently.
However, the end-user device host computer does not, necessarily, need to use the same IP address every day. Thinking back to the example of your favorite electronics store, you could move to a new apartment every week, but you still know where the electronics store is. The workers at the store do not need to know where you live. Likewise, servers typically don’t care that your PC has a different IP address today than yesterday.
As computers become smaller and lighter, it is more common for them to move from one network to another, where they would require a different IP address and network and or subnetwork. Actually, we find that in the course of a normal business day, many laptops and palmtops could move from one network to another many times per day. In reality, we find that on a worldwide basis, DHCP is used for everything from assigning IP addresses for corporate networks with thousands of hosts, to enabling a home Internet access router to automatically provide the correct Internet configuration to a single user’s computer.
DHCP defines the protocols used to allow computers and other devices to request a permanent IP address or lease a temporary address. The DHCP process is modeled on the client-server configuration. The DHCP process uses a server that maintains a list of pools of IP addresses available in each subnet. The user devices, or DHCP clients, can send the DHCP server a message, asking to be permanently assigned, or to lease an IP address. The DHCP server then suggests an IP address and, if accepted by the DHCP client, the server notes that the IP address is no longer available for assignment to any other hosts and the client has an address to use.
When this dynamic DHCP IP address allocation is used, the network administrator must configure the parameters on the DHCP server to control how leases are assigned, managed and, most importantly, how long the administrator wants the client leases to last. The administrator can set the lease time to infinity or, in essence, a permanent lease. However, administrators can choose a number that best suits the network. Some of the most often-used lease periods are:
- One Hour or Less
- One Day
- Three Days
- One Week
- One Month
- Three Months
- One Year
It’s interesting to note that with the DHCP process, clients no longer own their IP addresses. They lease them from the true owner, the DHCP server. As such, the primary job of both the DHCP server and the administrator of that server is to maintain and manage client addresses.
When properly configured, a DHCP server can also supply additional information to a client. For example, in addition to learning their IP address, a host needs to know the subnet mask to use, plus what default gateway to use, as well as the IP address(es) of any DNS servers. In most networks today, DHCP supplies all of these facts to a typical end-user host.
Author: David Stahl
