DHCP Architecture

Applies To: Windows Server 2008

The DHCP architecture consists of DHCP clients, DHCP servers, and DHCP relay agents on a network. The clients interact with servers using DHCP messages in a DHCP conversation to obtain and renew IP address leases.

DHCP client functionality

A DHCP client is any network-enabled device that supports the ability to communicate with a DHCP server in compliance with RFC 2131, for the purpose of obtaining dynamic leased IP configuration and related optional information.

DHCP provides support for client computers running any of the following operating systems:

• Windows Vista

• Windows Server 2008

• Windows XP

• Windows Server 2003

• Windows 2000

• Windows 98

• Windows Millennium Edition

• Windows NT version 4.0

Automatic IP configuration

DHCP supports Automatic Private IP Addressing (APIPA), which enables computers running Windows 2000, Windows XP, Windows Server 2003, Windows Vista, and Windows Server 2008 to configure an IP address and subnet mask if a DHCP server is unavailable at system startup and the Automatic private IP address alternate configuration setting is selected. This feature is useful for clients on small private networks, such as a small-business office or a home office.

On a computer running any of the previously mentioned Windows operating systems, the DHCP Client service uses the following process to auto-configure the client:

1. The DHCP client attempts to locate a DHCP server and obtain an IP address and configuration.

2. If a DHCP server cannot be found or does not respond after one minute, the DHCP client checks the settings on the Alternate Configuration tab of the properties of the TCP/IP protocol.

   If Automatic private IP address is selected, the DHCP client auto-configures its IP address and subnet mask by using a selected address from the Microsoft-reserved Class B network, 169.254.0.0, with the subnet mask 255.255.0.0. The DHCP client tests for an address conflict to ensure that the IP address is not in use on the network. If a conflict is found, the client selects another IP address. The client retries auto-configuration up to 10 times.

   If User Configured is selected, the DHCP client configures a static IP address configuration. The DHCP client tests for an address conflict to ensure that the IP address is not already in use on the network. If a conflict is found, the DHCP client indicates the error condition to the user.

3. When the DHCP client succeeds in selecting an address, it configures its network interface with the IP address. The client then continues to check for a DHCP server in the background every five minutes. If a DHCP server responds, the DHCP client abandons its selected IP address and uses the address offered by the DHCP server (and any other DHCP option information that the server provides) to update its IP configuration settings.

If the DHCP client obtained a lease from a DHCP server on a previous occasion, and the lease is still valid (not expired) at system startup, the client tries to renew its lease. If, during the renewal attempt, the client fails to locate any DHCP server, it attempts to ping the default gateway listed in the lease, and proceeds in one of the following ways:

• If the ping is successful, the DHCP client assumes that it is still located on the same network where it obtained its current lease, and continues to use the lease as long as the lease is still valid. By default the client then attempts, in the background, to renew its lease when 50 percent of its assigned lease time has expired.

• If the ping fails, the DHCP client assumes that it has been moved to a network where a DHCP server is not available. The client then auto-configures its IP address by using the settings on the Alternate Configuration tab. When the client is auto-configured, it attempts to locate a DHCP server and obtain a lease every five minutes.

Local storage

Windows Server 2008 DHCP supports local storage, which allows clients to store DHCP information on their own hard disks. Local storage is useful because it enables the client to store its last leased IP address, so that when the client starts, it first attempts to renew the lease of its previous IP address. Local storage also enables a client to be shut down and restarted and it will use its previously leased address and configuration, even if the DHCP server is unreachable or offline at the time that the client computer is restarted.

DHCP server responsibilities

The DHCP servers maintain scopes, reservations, and options as set by the administrator.

Scopes

A scope must be properly defined and activated before DHCP clients can use the DHCP server for automatic TCP/IP configuration. A DHCP scope is an administrative collection of IP addresses and TCP/IP configuration parameters that are available for lease to DHCP clients of a specific subnet. The network administrator creates a scope for each subnet.

A scope has the following properties:

• A scope name, assigned when the scope is created.

• A range of possible IP addresses from which to include or exclude addresses used in DHCP lease offers.

• A unique subnet mask, which determines the network ID for an IP address in the scope.

• Lease duration values.

Each DHCP scope can have a single continuous range of IP addresses. To use several address ranges within a single scope, you must first define the entire address range for the scope, and then set exclusion ranges.

Lease durations

When a scope is created, the lease duration is set to eight days by default. However, there are situations when the administrator might want to change the lease duration. The following are examples of adjusting the lease duration due to individual network consideration:

• An organization has a large number of IP addresses available and client configurations rarely change. The administrator increases the lease duration to reduce the frequency of lease renewal exchanges between clients and the DHCP server. Because the DHCP clients are renewing their leases less frequently, DHCP-related network traffic is reduced.

• An organization has a limited number of IP addresses available and client configurations change frequently or clients move often in or out of the network. The administrator reduces the lease duration to increase the rate at which unused addresses are returned to the available address pool for reassignment.

For example, consider the ratio between connected computers and available IP addresses. If 40 computers share 254 available addresses, the demand for reusing addresses is low. A long lease time, such as a few months, might be appropriate in such a situation. However, if 230 computers must share the same address pool, demand for available addresses is higher, and a shorter lease time (for example, a few days), is more appropriate.

Exclusion ranges

When you create a new scope, immediately exclude the addresses of existing statically configured computers. By using exclusion ranges, you can exclude specific IP address ranges within a scope so that those addresses are not offered to clients. Assign IP addresses within exclusion ranges to computers or devices that must have a static IP address, such as servers, firewalls, or routers.

You can use excluded IP addresses on your network by manually configuring these addresses at computers that do not use DHCP to obtain an address, or by configuring reservations for these addresses.

Reservations

You can reserve IP addresses for assignment to specified computers or devices on the network. Reservations ensure that a specified hardware device on a subnet always receives the same IP address lease. Use reservations for DHCP-enabled devices that must always have the same IP address on your network, such as servers that do not support Domain Name System (DNS) dynamic update.

Superscopes

A superscope allows a DHCP server to provide leases from more than one scope to clients on a single physical subnet. Before you can create a superscope, you must use the DHCP Microsoft Management Console (MMC) snap-in to define at least one of the scopes to be included in the superscope. Scopes added to a superscope are called member scopes. Superscopes can resolve DHCP Server service issues in several different ways; these issues include situations in which:

• Support is needed for DHCP clients on a single physical network segment — such as a single Ethernet local area network (LAN) segment — where multiple logical IP networks are used. When more than one logical IP network is used on a physical network, these configurations are also known as multinets. In a situation where multinets are used, clients might not be able to communicate directly with each other, because the clients might be on different logical subnets, even if they are on the same physical network segment. In this case, routing must be enabled to allow the clients to communicate with each other. Also, a router or BOOTP/DHCP relay agent must be configured on the subnet to allow DHCP messages to travel between the logical subnets.

• Support is needed for DHCP clients that are in a multinet located on the other side of BOOTP relay agents.

• Clients need to be migrated to a new scope.

Interactions between client and server

DHCP servers and DHCP clients communicate through a series of DHCP messages. To obtain a lease, the DHCP client initiates a conversation with a DHCP server using a series of these DHCP messages.

DHCP messages

The following list includes the eight types of messages that can be sent between DHCP clients and servers.

DHCPDiscover

Broadcast by a DHCP client when it first attempts to connect to the network. The DHCPDiscover message requests IP address information from a DHCP server.

DHCPOffer

Broadcast by each DHCP server that receives the client DHCPDiscover message and has an IP address configuration to offer to the client. The DHCPOffer message contains an unleased IP address and additional TCP/IP configuration information, such as the subnet mask and default gateway. More than one DHCP server can respond with a DHCPOffer message. The client accepts the best offer, which, for a Windows DHCP client, is the first DHCPOffer message that it receives.

DHCPRequest

Broadcast by a DHCP client after it selects a DHCPOffer. The DHCPRequest message contains the IP address from the DHCPOffer that it selected. If the client is renewing or rebinding to a previous lease, this packet might be unicast directly to the server.

DHCPAck

Broadcast by a DHCP server to a DHCP client acknowledging the DHCPRequest message. At this time, the server also forwards any options. Upon receipt of the DHCPAck, the client can use the leased IP address to participate in the TCP/IP network and complete its system startup. This message is typically broadcast, because the DHCP client does not officially have an IP address that it can use at this point. If the DHCPAck is in response to a DHCPInform, then the message is unicast directly to the host that sent the DHCPInform message.

DHCPNack

Broadcast by a DHCP server to a DHCP client denying the client’s DHCPRequest message. This might occur if the requested address is incorrect because the client moved to a new subnet or because the DHCP client’s lease has expired and cannot be renewed.

DHCPDecline

Broadcast by a DHCP client to a DHCP server, informing the server that the offered IP address is declined because it appears to be in use by another computer.

DHCPRelease

Sent by a DHCP client to a DHCP server, relinquishing an IP address and canceling the remaining lease. This is unicast to the server that provided the lease.

DHCPInform

Sent from a DHCP client to a DHCP server, asking only for additional local configuration parameters; the client already has a configured IP address. This message type is also used by DHCP servers running Windows Server 2008 to detect unauthorized DHCP servers.

DHCP lease process

A DHCP-enabled client obtains a lease for an IP address from a DHCP server. Before the lease expires, the DHCP client must renew the lease or obtain a new lease. Leases are retained in the DHCP server database for a period of time after expiration. By default, this grace period is four hours and cleanup occurs once an hour for a DHCP server running Windows Server 2008. This protects a client’s lease in case the client and server are in different time zones, the internal clocks of the client and server computers are not synchronized, or the client is off the network when the lease expires.

Obtaining a new lease

A DHCP client initiates a conversation with a DHCP server when it is seeking a new lease, renewing a lease, rebinding, or restarting. The DHCP conversation consists of a series of DHCP messages passed between the DHCP client and DHCP servers. The following figure shows an overview of this process when the DHCP server and DHCP client are on the same subnet.

DHCP Lease Process Overview

1. The DHCP client requests an IP address by broadcasting a DHCPDiscover message to the local subnet.

2. The client is offered an address when a DHCP server responds with a DHCPOffer message containing an IP address and configuration information for lease to the client. If no DHCP server responds to the client request, the client sends DHCPDiscover messages at intervals of 0, 4, 8, 16, and 32 seconds, plus a random interval of between -1 second and 1 second. If there is no response from a DHCP server after one minute, the client can proceed in one of two ways:

   • If the client is using the APIPA alternate configuration, the client self-configures an IP address for its interface.

   • If the client does not support alternate configuration, such as APIPA, or if IP auto-configuration has been disabled, the client network initialization fails.

   In both cases, the client begins a new cycle of DHCPDiscover messages in the background every five minutes, using the same intervals as before (0, 4, 8, 16, and 32 seconds), until it receives a DHCPOffer message from a DHCP server.

3. The client indicates acceptance of the offer by selecting the offered address and broadcasting a DHCPRequest message in response.

4. The client is assigned the address and the DHCP server broadcasts a DHCPAck message in response, finalizing the terms of the lease.

When the client receives acknowledgment, it configures its TCP/IP properties by using the DHCP option information in the reply, and completes its initialization of TCP/IP.

In rare cases, a DHCP server might return a negative acknowledgment to the client. This can happen if a client requests an address that is not valid or a duplicate. If a client receives a negative acknowledgment (DHCPNack), the client must begin the entire lease process again.

When the DHCP client and the DHCP server are on the same IP broadcast subnet, the DHCPDiscover, DHCPOffer, DHCPRequest, and DHCPAck messages are sent to identify clients by means of IP-level broadcasts sent to the limited broadcast address and the media access control (MAC) broadcast address.

When the DHCP server and DHCP client are not on the same subnet, either a router or a host on the DHCP client’s subnet must act as a DHCP relay agent to support the forwarding of DHCP messages between the DHCP client and the DHCP server.

Renewing a lease

The DHCP client first attempts to renew its lease when 50 percent of the original lease time, known as T1, has passed. At this point, the DHCP client sends a unicast DHCPRequest message to the DHCP server that originally granted its lease. If the server is available, and the lease is still available, the server responds with a unicast DHCPAck message, and the lease is renewed.

If the original DHCP server is available, but the client’s current lease is no longer available, the DHCP server responds with a DHCPNack message, and the client immediately starts the process to obtain a new lease. This can happen if the client has changed subnets or if the DHCP server cannot fulfill the lease request for some other reason.

If there is no response from the DHCP server, the client waits until 87.5 percent of the lease time, known as T2, has passed. At T2, the client enters the rebinding state, and broadcasts a DHCPRequest message to attempt to renew the lease from any available DHCP server. If no DHCP server is available by the time the lease expires, the client immediately unbinds itself from the existing lease and starts the process to obtain a new lease, beginning with a DHCPDiscover message.

Preventing address conflicts

Windows Server 2008 DHCP has both server-side and client-side conflict detection to prevent duplicate IP addresses on your network.

Client conflict detection

Client computers running Windows Server 2008, Windows Vista, Windows Server 2003, Windows XP, Windows 2000, Windows NT 4.0, Windows Millennium Edition, and Windows 98 automatically check to determine if an IP address is already in use before using it.

After the DHCP client receives a lease from the DHCP server, the client sends an Address Resolution Protocol (ARP) request to the address that it has been assigned. If a reply to the ARP request is received, the client has detected a conflict and sends a DHCPDecline message to the DHCP server. The DHCP server attaches a BAD_ADDRESS value to the IP address in the scope for the length of the lease. The client then begins the lease process again, and is offered the next available address in the scope.

Server conflict detection

If your network includes older DHCP clients that do not perform conflict detection themselves, you can enable conflict detection on the DHCP server. By default, the Windows Server 2008 DHCP Server service does not perform any conflict detection.

To detect conflicts, the DHCP server pings (sends an ICMP Echo message to) an IP address before offering that address to clients in a new lease. The DHCP server only pings addresses that have not been successfully and previously leased. If a client requests a lease on an IP address that it already had or is requesting a renewal, the DHCP server does not ping the IP address.

If conflict detection is enabled, an administrator-defined number of pings are sent. The server waits 1 second for a reply. Because the time required for a client to obtain a lease is equal to the number of pings used, choose this value carefully because it directly impacts the overall performance of the server. In general, one ping is sufficient.

If a response to the ping is received, a conflict is registered and that address is not offered to clients requesting a lease from the server. The DHCP server then attaches a BAD_ADDRESS value to that IP address in the scope. The DHCP server then tries to lease the next available address. If the duplicate address is removed from the network, the BAD_ADDRESS value attached to the IP address can be deleted from the scope’s list of active leases, and then the address returns to the pool. Addresses are marked as BAD_ADDRESS for the length of the lease for which the scope is configured. If the BAD_ADDRESS entry is not manually removed, it will automatically be removed after a period of time equal to the lease time for the scope.

DHCP options

DHCP options are additional configuration parameters that a DHCP server assigns to clients. Options can also be used for DHCP communication between the server computer and client computers.

The most specific options take precedence over the least specific options. This simplifies DHCP management and allows a flexible administration that can range from per-server default settings to common settings for a specific subnet and individualized client settings when needed for special circumstances. In most cases, the option values are specified in the Options dialog box on the DHCP server, scope, or reservation.

DHCP options can be configured for specific values and enabled for assignment and distribution to DHCP clients based on:

• Server options. These options apply globally for all scopes and classes defined at each DHCP server and any clients that it services. Configured server option values always apply unless they are overridden by options assigned to other scope, class, or client reservation.

• Scope options. These options apply to any clients that obtain a lease within that particular scope. Configured scope option values always apply to all computers obtaining a lease in a given scope unless they are overridden by options assigned to class or client reservation.

• Class options. These options apply to any clients that specify that particular DHCP Class ID value when obtaining a scope lease. Configured class option values always apply to all computers configured as members in a specified DHCP option class unless they are overridden by options assigned to a client reservation.

• Reserved client options. These options apply only to the client corresponding to the reservation. Reserved client option values override all other server, scope, or class assigned option values.

Options are typically applied at each DHCP server at the server or scope level. To precisely manage or customize option settings for a group or class of computers, specify either a user or vendor class assignment that overrides the broader server or scope option defaults.

For special requirements, such as clients with special functions, assign options for specific reserved clients.

Options can also be used to separate and distribute appropriate options for clients with similar or special configuration needs. For example, DHCP clients on the same floor of a building can be configured with the same DHCP Class ID value to assign them membership in the same option class. You can then distribute additional or varied option data to that class during the lease process, overriding any scope or globally provided default options.

Many options are predefined on a DHCP server running Windows Server 2008. Other standard DHCP options can be added as needed to support DHCP client software that recognizes or requires them. The DHCP Server service running on Windows Server 2008 supports all options defined in RFC 2132, although most DHCP clients use or support only a small subset of the available RFC-specified options.

The following table contains a list of default DHCP options requested by DHCP clients running Windows Server 2008, Windows Vista, Windows Server 2003, and Windows XP. For a complete reference of DHCP options, see DHCP Tools and Options.

Default DHCP options

DHCP option parameters

DHCP servers can be configured to provide optional data that fully configures TCP/IP on a client. Some of the most common DHCP option types configured and distributed by the DHCP server during the lease process include parameters for the default gateway, DNS, and WINS.

Clients can be configured with:

• Information options. You can explicitly configure these options and any associated values provided to clients.

• Protocol options. You can implicitly configure these options used by the DHCP Server service based on server and scope property settings.

You can use the DHCP snap-in to configure these properties and set them for an entire scope or for a single, reserved client scope.

Information options

The following table lists the most common types of DHCP information options that can be configured for DHCP clients. These options can be enabled and configured for each scope that you configure on a DHCP server. Depending on your network infrastructure, some of these options can be configured as server options, such as DNS domain name.

Common information options

Clients can request these DHCP options, and can use the values to set their TCP/IP configurations for the duration of the lease.

Protocol options

The following table shows protocol options that DHCP clients can be configured to use when communicating with a DHCP server to obtain or renew a lease.

Common protocol options

The values provided to clients for lease time, T1, and T2 are taken from the scope settings on the DHCP server. The value provided for DHCP message type is automatically set, depending on which packet of the DHCP conversation is being sent.

Option classes

Option classes allow quick introduction of custom applications for enterprise networks. DHCP option classes provide a way to easily configure network clients with the parameters required for custom applications. Equipment from multiple vendors on a network can also use different option code numbers for different functions. The options used to support vendor classes — the vendor class identifier and the vendor-specific option — are defined in the Internet DHCP options standard reference, RFC 2132.

Windows Server 2008 includes two types of option classes: vendor-defined and user-defined. These classes can be configured on your servers to offer specialized client support in the following ways:

• Add and configure vendor-defined classes for managing DHCP options assigned to clients identified by vendor type.

• Add and configure user-defined classes for managing DHCP options assigned to clients that need a similar DHCP option configuration.

After options classes are defined on a DHCP server, scopes on the server can be configured to assign options for specific user-defined and vendor-defined option classes.

Vendor classes

Vendor-defined option classes can be used by DHCP clients to identify the client’s vendor type and configuration when obtaining a lease from the DHCP server. The client can include the vendor class ID option (option code 60) when it requests or selects a lease from a DHCP server to identify its vendor class during the lease process.

The vendor class identifier information is a string of character data interpreted by the DHCP servers. Vendors can choose to define specific vendor class identifiers to convey particular configuration or other identification information about a client. For example, the identifier might encode the client’s hardware or software configuration. Most vendor types are derived from standard reserved hardware and operating system-type abbreviation codes listed in RFC 1700.

When vendor options are specified, the server performs the following additional steps to provide a lease to the client:

• The server verifies that the vendor class identified by the client request is a recognized class defined on the server.

  If the vendor class is recognized, the server checks to see if any additional DHCP options are configured for this class in the active scope.

  If the vendor class is not recognized, the server ignores the vendor class identified in the client request, and returns options allocated to the default vendor class (which includes all DHCP Standard options).

• If the scope contains options configured specifically for use with clients in this vendor-defined class, the server returns those options using the vendor-specific option type (option code 43) as part of its acknowledgment message.

In most cases, the default vendor class — the DHCP Standard option class — provides a default vendor class for any Windows DHCP clients or other DHCP clients that do not specify a vendor class ID. In some cases, you might define additional vendor classes for other DHCP clients, such as printers or some types of UNIX clients. When you add other vendor classes for these purposes, make sure that the vendor class identifier you use to configure the class at the server matches the identifier used by clients for your other, non-Microsoft vendor.

User classes

User classes allow DHCP clients to differentiate themselves by specifying which type of client they are, such as desktop or server computer. For computers running Windows Server 2008, Windows Vista, Windows Server 2003, Windows XP, and Windows 2000, you can define user class identifiers to convey information about a client’s software configuration, its physical location in a building, or its user preferences. For example, an identifier can specify that DHCP clients are members of a user class called “2nd floor, West”, which needs a specific set of router, DNS, and WINS server settings. An administrator can then configure the DHCP server to include different option values, depending on the user class of client receiving the lease.

Windows Server 2008 user classes can be used as follows:

• DHCP client computers can include the DHCP user class option when sending DHCP request messages to the DHCP server. This can specifically identify the client as part of a user class on the server.

• DHCP servers running the Windows 2000 Server, Windows Server 2003, or Windows Server 2008 DHCP Server service can recognize and interpret the DHCP user class option from clients and provide additional options (or a modified set of DHCP options) based on the client’s user class identity.

For example, shorter leases can be assigned to wireless clients. Or perhaps a particular set of clients might need a specific set of routes, a specific DNS server, or a specific default gateway.

A user class can be either a default or custom user class. Microsoft provides three default user classes, as described in the following table.

Default user classes provided by Microsoft

Use the Microsoft default user classes to isolate specific configuration details for clients with special needs, such as older clients or clients that use BOOTP or Routing and Remote Access. For example, you might want to include and assign special BOOTP option types (such as option codes 66 and 67) for clients that are BOOTP type, or shorten the lease time for remote access clients.

You can also add and configure custom user classes for use by DHCP clients running Windows 2000, Windows XP, Windows Server 2003, Windows Vista, and Windows Server 2008. For a custom user class to work properly, the client must use the same custom identifier when requesting options as was used when the class was defined on the DHCP server.

The user class option field allows only one ASCII text string to be used for identifying clients. This means each client computer can be identified only as a member of a single user class by the DHCP server. You can use additional user classes to make new hybrids from your other user classes to accommodate clients that need configuration for multiple user classes. For example, if you have two user classes, one called “mobile” with short lease times assigned and another called “engineer” with an option assigned to configure a high-performance server for its clients, you can make a new hybrid user class called “mobile-engineer” that contains both special option value settings.

MADCAP and multicast DHCP

Multicast Address Dynamic Client Allocation Protocol (MADCAP) is modeled after the DHCP standard. MADCAP assists in simplifying and automating configuration of multicast groups on your network, but it is not required for the operation of multicast groups or for the DHCP Server service. Multicast scopes provide only multicast address configuration and do not support or use other DHCP-assignable options.

Multicast scopes configured on the DHCP server define ranges of IP multicast addresses. Similar to allocating unicast IP addresses, IP multicast addresses are allocated to MADCAP clients. A MADCAP address is configured separately from a primary IP address. Computers that use either static or dynamic IP configuration through a DHCP server can be MADCAP clients.

In Windows Server 2008, the DHCP Server service supports both DHCP and MADCAP, although these services function separately. Clients of one do not depend on the use or configuration of the other.

Clients that do not support the MADCAP service or are unable to contact and obtain multicast configuration from a MADCAP server can be configured in other ways so that they participate in either permanent or temporary multicast groups on the network.

In all TCP/IP networks, each computer requires a unique primary unicast IP address for each network interface. You must assign this required primary unicast IP address before you can configure a computer to support and use secondary IP addresses such as multicast IP addresses.