Sample DHCP User

Typical DHCP Administration

Leases

Scopes and Policies

After Beth’s workstation (bethpc) is configured with DHCP, these actions occur when she first starts up:

1. Her pc automatically requests an IP address from a DHCP server on the network.
2. The DHCP server offers her a lease that is an IP address with the configuration data necessary to use the Internet. Nobody else uses the leased address, and it is valid only for her pc.
3. Before the address lease expires, bethpc renews it, thereby extending the expiration time. It continues to use the lease right up to its expiration or if it cannot reach the server.
4. If Beth relocates to another department and her pc moves to a different subnet, her current address expires and becomes available for others. When Beth starts her pc at its new location, it leases an address from an appropriate DHCP server on the subnet (see the image below).

As long as the DHCP server has the correct configuration data, none of the workstations or servers using DHCP will ever be configured incorrectly. Therefore, there is less chance of incurring network problems from incorrectly configured devices and servers that are difficult to trace.

The example shows the DHCP protocol with a set of DHCP servers that provide addresses on different subnets. To further simplify the administration of address pools, network routers are often configured as DHCP relay agents to forward client messages to a central DHCP server. This server is configured with address pools for a group of subnets.

To use DHCP, you must have at least one DHCP server on the network. After you install the server:

• Define a scope of IP addresses that the DHCP server can offer to DHCP clients. You no longer need to keep track of which addresses are in use and which are available.
• Configure a secondary server to share the distribution or handle leases if the first DHCP server goes down. This is known as DHCP failover. For information on Managing DHCP Failover, see Managing DHCP Failover.

One of the most significant benefits of DHCP is that it can dynamically configure devices with IP addresses and associate leases with the assigned addresses. DHCP uses a lease mechanism that offers an automated, reliable, and safe method for distributing and reusing addresses in networks, with little need for administrative intervention. As system administrator, you can tailor the lease policy to meet the specific needs of your network.

Leases are grouped together in an address pool, called a scope, which defines the set of IP addresses available for requesting hosts. A lease can be reserved (the host always receives the same IP address) or dynamic (the host receives the next available, unassigned lease in the scope). The DHCP server of the site is configured to lease addresses 192.168.1.100 through 192.168.1.199 (see the image below).

If you plan not to have more network devices than configured addresses for the scope, you can define long lease times, such as one to two weeks, to reduce network traffic and DHCP server load.

A scope contains a set of addresses for a subnet, along with the necessary configuration parameters. You must define at least one scope for each subnet for which you want dynamic addressing.

A policy includes lease times and other configuration parameters that a DHCP server communicates to clients. Use policies to configure DHCP options that the DHCP server supplies to a client upon request. Policies ensure that the DHCP server supplies all the correct options for scopes without having to do so separately for each scope (see the image below).

The difference between scopes and policies is that scopes contain server information about addresses, such as which address is leasable and whether to ping clients before offering a lease. Policies contain client configuration data, such as the lease duration and address of the local DNS server.

Policies are especially useful if you have multiple scopes on a server. You can create policies that apply to all or selected scopes. The Cisco Prime Network Registrar policy hierarchy is a way to define policies from least to most specific. For example, you usually specify a router option for each policy, which means that you would need a policy for each scope. Scope-specific policies like this can be defined in a scope-embedded policy. More general policies, such as those referring to lease times, can be applied in a system-wide policy (see Configuring DHCP Policies). You can also write extensions to handle policy assignments (see Using Extensions to Affect DHCP Server Behavior).

Determining Links and Prefixes

Generating Addresses

Generating Delegated Prefixes

Prefix Stability

Virtual Private Networks

Subnet Allocation and DHCP Address Blocks

Virtual private networks (VPNs) allow the possibility that two pools in separate networks can have the same address space, with these two pools having overlapping private network addresses. This can save address resources without having to use valuable public addresses. These VPN addresses, however, require a special designator to distinguish them from other overlapping IP addresses. Cisco Prime Network Registrar DHCP servers that are not on the same VPN as their clients can now allocate leases and addresses to these clients, and can distinguish the addresses from one VPN to another.

Through changes made to the Cisco Prime Network Registrar DHCP server and Cisco IOS DHCP Relay Agent, the DHCP server can service clients on multiple VPNs. A VPN distinguishes a set of DHCP server objects, making them independent of otherwise identical objects in other address spaces. You can define multiple VPNs containing the same addresses. You create a VPN based on the VPN identifier configured in the Cisco IOS Relay Agent.

The illustration below shows a typical VPN-aware DHCP environment. The DHCP Relay Agent services two distinct VPNs, blue and red, with overlapping address spaces. The Relay Agent has the interface address 192.168.1.1 on VPN blue and is known to DHCP Server 1 as 172.27.180.232. The server, which services address requests from DHCP Client 1 in VPN blue, can be on a different network or network segment than the client, and can be in a failover configuration with DHCP Server 2 (see Managing DHCP Failover). The Relay Agent can identify the special, distinguished route of the client address request to the DHCP server, as coordinated between the Relay Agent and Cisco Prime Network Registrar administrators. The DHCP servers can now issue leases based on overlapping IP addresses to the clients on both VPNs.

Effect on DNS of Obtaining Leases

Effect on DNS of Releasing Leases

Effect on DNS of Reacquiring Leases

For ExampleCo, the administrator creates a scope on the DHCP server and allocates 100 leases (192.168.1.100 through 192.168.1.199). Each device gets its owner name. The administrator also configures the DHCP server to use DNS update and associates it with the correspondingly configured DNS server. The administrator does not need to enter the names in the DNS server database.

Monday morning, Beth (user of bethpc) tries to log into a website without having an address. When her host starts up, it broadcasts an address request (see the image below).

The DHCP server then:

1. Gives bethpc the next available (unassigned) IP address (192.168.1.125).
2. Updates her DNS server with the hostname and address (bethpc 192.168.1.125).

Beth can now access the website. In addition, programs that need to translate the name of Beth’s machine to her IP address, or the other way around, can query the DNS server.

When Beth returns from her trip to start up her host again:

1. Her pc broadcasts for an IP address.
2. The DHCP server checks if the host is on the correct network. If so, the server issues an address. If not, the server on the correct network issues the address.
3. The DHCP server updates the DNS server again with the host and address data.

Later that day, Beth learns that she needs to travel out of town. She turns off her host, which still has a leased address that is supposed to expire after three days. When the lease is released, the DHCP server:

1. Acknowledges that the IP address is now available for other users (see the figure below).
2. Updates the DNS server by removing the hostname and address. The DNS server no longer stores data about bethpc or its address.

In order to keep the failover pair operating in spite of a network partition, in which both can communicate with clients but not with each other, you must make available more addresses than the addresses needed to run a single server. Configure the main server to allocate a percentage of the currently available (unassigned) addresses in each scope or prefix delegation address pool to its partner. These addresses become unavailable to the main server. The partner uses them when it cannot talk to the main server and does not know if it is down. However, when the failover partners are in communication, they periodically rebalance these pools.

The backup server needs enough addresses from each scope or prefix to satisfy the requests of all new DHCP clients that arrive during the period in which the backup does not know if the main server is down. In Cisco Prime Network Registrar 8.2 or later, the default backup percentage for a failover pair is 50%. This ensures that during the failover the other partner has equal number of addresses.

Even during PARTNER-DOWN state, the backup server waits for the lease expiration and the maximum client lead time (MCLT), a small additional time buffer, before reallocating any leases. When these times expire, the backup server offers:

• Leases from its private pool of addresses.
• Leases from the main server pool of addresses.
• Expired leases to new clients.

During the working hours, if the administrative staff can respond within two hours to a COMMUNICATIONS INTERRUPTED state to determine if the main server is working, the backup server needs enough addresses to support a reasonable upper bound on the number of new DHCP clients that might arrive during those two hours.

During off-hours, if the administrative staff can respond within 12 hours to the same situation, and considering that the arrival rate of previously unheard from DHCP clients is also less, the backup server then needs enough addresses to support a reasonable upper bound on the number of DHCP clients that might arrive during those 12 hours.

Consequently, the number of addresses over which the backup server requires sole control would be the greater of the numbers of addresses given out during peak and non-peak times, expressed as a percentage of the currently available (unassigned) addresses in each scope or prefix.

Note	Starting in 8.2, the default use-safe-period is enabled for the DHCP failover pair and the default safe period is 4 hours. This ensures that if the failover partner is in COMMUNICATIONS-INTERRUPTED state for 4 hours, it will enter PARTNER-DOWN state automatically after the safe period elapses.

DHCP Processing Without Client-Classes

DHCP Processing with Client-Classes

Defining Scopes for Client-Classes

Choosing Networks and Scopes

To understand how you can apply client-class processing, it is helpful to know how the DHCP server handles client requests. The server can perform three tasks:

• Assign an IP address.
• Assign the appropriate DHCP options (configuration parameters).
• Optionally assign a fully qualified domain name (FQDN) and update the DNS server with that name.

The DHCP server:

1. Assigns an address to the client from a defined scope—To choose an address for the client, the DHCP server determines the client subnet, based on the request packet contents, and finds an appropriate scope for that subnet.

   If you have multiple scopes on one subnet or several network segments, which is known as multinetting, the DHCP server may choose among these scopes in a round-robin fashion, or you can change the priority of the scope choice by using the DHCP server address allocation priority feature (see Configuring Multiple Scopes Using Allocation Priority). After the server chooses a scope, it chooses an available (unassigned) address from that scope:

   1. It assigns DHCP option values from a defined policy. Cisco Prime Network Registrar uses policies to group options. There are two types of policies: scope-specific and system default. For each DHCP option the client requests, the DHCP server searches for its value in a defined sequence.

   2. If the scope-specific policy contains the option, the server returns its value to the client and stops searching.
   3. If not found, the server looks in the system default policy, returns its value, and stops searching.
   4. If neither policy contains the option, the server returns no value to the client and logs an error.
   5. The server repeats this process for each requested option.
2. With DNS update in effect, the server assigns an FQDN to the client. If you enabled DNS update, Cisco Prime Network Registrar enters the client name and address in the DNS host table. See DNS Update. The client name can be:
   • Its name as specified in the client lease request (the default value).
   • Its MAC address (hardware address; for example, 00:d0:ba:d3:bd:3b).
   • A unique name using the default prefix dhcp or a specified prefix.

When you enable the client-class facility for your DHCP server, the request processing performs the same three tasks of assigning IP addresses, options, and domain names as described in DHCP Processing Without Client-Classes, but with added capability. The DHCP server:

1. Considers the client properties and client-class inclusion before assigning an address —As in regular DHCP processing, the DHCP server determines the client subnet. The server then checks if there is a client-class defined or a MAC address for this client in its database. If there is:
   1. A client-class defined by a client-class lookup ID expression, the client is made a member of this client-class.
   2. No MAC address, it uses the default client. For example, the default client could have its client-class name set to Guest, and that client-class could limit (using options and address selection) what network operations such clients are permitted.
   3. No MAC address and no default client, the server handles the client through regular DHCP processing.
   4. No client-specifier, but a MAC address, the MAC address is converted into a client-specifier. An unknown client is mapped to the default client, if the default client is defined.

   The scopes must have addresses on client-accessible subnets. That is, they must have a selection tag that associates them with a client-class. To assign the same clients to different address pools, you must use separate scopes.

   For example, a scope would either have a selection tag of Employee or Guest, but not both. In this case, there are two scopes for each subnet; one with the selection tag Employee, and the other with Guest. Each scope has a different associated policy and address range that provides the appropriate access rights for the user group.

2. Checks for client-class DHCP options —In regular DHCP processing, the server checks the scope-specific and system default DHCP options. With client-class, it also first checks the client-specific and client-class-specific options.
3. Provides additional FQDN assignment options —Beyond the usual name assignment process of using the hostname the client requests, the server can:
   • Provide an explicit hostname that overrides it.
   • Drop the client-requested hostname and not replace it.
   • Synthesize a hostname from the client MAC address.

The motivating factor for using client-classes is often to offer an address from one or another address pool to a client. Another motivating factor might be to provide clients with different option values or lease times. Offering clients addresses from separate pools requires defining more than one scope.

To get more than one scope on a subnet, they must come from the same network segment. Networks are not configured directly in Cisco Prime Network Registrar, but are inferred from scope configurations. Scopes become related (end up in the same network):

• Implicitly —Two scopes have the same network number and subnet mask. These scopes naturally end up on the same network without explicit configuration.
• Explicitly —One scope is marked as a secondary to another. This is required when the scope marked as a secondary has a network and subnet mask unrelated to the primary. An example is putting a set of 10.0.0.0 network addresses on a normal, routable network segment.

When the Cisco Prime Network Registrar DHCP server reads the scope configuration from its database, it places every scope in a network, and logs this information. Scopes with the same network number and subnet mask end up on the same network, while a secondary scope ends up on the primary scope network.

When a DHCP packet arrives, the server determines the address from which it came by:

• When a DHCPv4 packet arrives the server determines the gateway address (giaddr), if there was one, for packets sent through a BOOTP relay.
• For information on DHCPv6, see Determining Links and Prefixes.
• Interface address of the interface on which the broadcast packet arrived, if the DHCP client is on a network segment to which the DHCP server is also directly connected.

In all cases, the DHCP server determines a network from the gateway or interface address. Then, if the network has multiple scopes, the server determines from which scope to allocate an address to the DHCP client. It always looks for a scope that can allocate addresses to this type of client. For example, a DHCP client needs a scope that supports DHCP, and a BOOTP client needs one that supports BOOTP. If the client is a DHCP client and there are multiple scopes that support DHCP, each with available (unassigned) addresses, the DHCP server allocates an IP address from any of those scopes, in a round-robin manner, or by allocation priority.

Selection tags and client-classes let you configure the DHCP server to allocate IP addresses from:

• One or more scopes on a network to one class of clients.
• A different set of scopes to a different class of clients.

In the latter case, the gateway or interface address determines the network. The client-class capability, through the mechanism of the selection tags, determines the scope on the network to use.