Making sure a DHCP server is always available is critical if your network uses dynamic TCP/IP addressing.
Microsoft DHCP server became much more popular in Windows 2000 environments, where it became part of the overall strategy for managing IP addressing, host namespace, and name resolution (due to its close integration with Microsoft's implementation of DNS). Because of its significance, it is imperative to have a solid plan that allows you to quickly recover from DHCP server failures.
One approach to ensuring DHCP server availability is to install multiple DHCP servers and divide the list of available IP addresses on each subnet into multiple ranges, one per server. In the simplest case of two DHCP servers, configure each with the scopes that have matching start and end address. Next, for each one create mutually exclusive exclusion lists. For example, if your network is using class C nonsubnetted network 192.168.168.0/24, then, on both servers, you should create the scope with the start IP address 192.168.0.1 and the end IP address 192.168.168.254. Your choice of exclusion lists depends on whether you want both servers to share the load equally or whether one of them will be a primary choice for your DHCP clients. For example, to balance the load, you would configure the range 192.168.168.1-192.168.168.127 on the first server and 192.168.168.128-18.104.22.168 on the second.
In order for this configuration to work, you have to ensure that broadcasts from DHCP clients will reach both servers. Typically, this is done either by installing DHCP relay agents on the servers that reside on clients subnet or by configuring routers as BOOTP Relay Agents.
In addition to providing redundancy, you should also ensure regular backups of the DHCP database. Fortunately, the backup takes place automatically by default. Its behavior is determined by Registry entries that reside in the following key:
The Registry entries contain the following values:
Determines the location of the backup (set initially to %SystemRoot%\System32\DHCP\Backup).
Determines the frequency of the automatic backup, in minutes (the default is 60).
Can be used to force the restore by using the existing backup (by setting RestoreFlag to 1). Typically, the operating system does this automatically if it detects the DHCP database corruption.
Windows also automatically backs up the content of the Registry key HKLM\SOFTWARE\Microsoft\DHCPServer\Configuration to the DHCPCFG file, which resides in the Backup folder.
Recovering the database involves restoring both the database files and the Registry settings. You should first stop the DHCP server and then copy the files and load the Registry hive (using REGEDT32.EXE) to their target location by overwriting the existing HKLM\SOFTWARE\Microsoft\DHCPServer\Configuration Registry key. After you have restored the database file, you should change the default of 0 conflict-detection attempts (from the Advanced tab of Server properties in the DHCP MMC console) to a nonzero value (5 is the maximum).
Another option is to use the NETSH command-line utility to back up and restore configuration of the DHCP server database. NETSH's functionality is provided through a number of helper DLLs, each dealing with a particular type of Windows networking component. NETSH allows you to dump the configuration of the DHCP server (including all superscopes, scopes, exclusion ranges, and reservations) into a text file that later can be used to restore. Note, however, that NETSH does not back up information about existing leases, which are stored in the DHCP database.
To create the DHCP configuration dump file, execute the following command, where IPAddressOrName is the IP address or name of your DHCP server (note that this command can be executed remotely):
NETSH DHCP SERVER IPAddressOrName DUMP > C:\DHCPCfg.txt
To restore the DHCP server configuration settings using the same file, run this command:
NETSH EXEC C:\DHCPCfg.txt