DHCP

Introduction

My network uses IPv4 addressing, using a single subnet supporting up to 254 devices on the network. In a more typical home environment, the local IP addresses would either be static, or dynamically assigned from an internet facing router using the Dynamic Host Configuration Protocol (DHCP). As well as assigning the client's network address, the DHCP server also supplies other network configuration parameters, such as Default Gateway and Domain Name System (DNS) server addresses.

Although I have a "professional" grade internet gateway/router on my network, a Draytek Vigor 2860Vn+, its DHCP server does not support all of the extended DHCP options that I need, such as TFTP server name that is needed to dynamically provision some of my IP phones. Therefore, my Windows Server 2008R2 Domain Controllers have the DHCP role configured and DHCP is disabled on my router.

IP Address Range		Type	Service
Low	High
.1	.100	Static	Infrastructure (servers, etc.)
.101	.180	DHCP	Consumer devices (PCs, etc.)
.181	.200	Static	VOIP (phones, gateway, server)
.201	.254	Static	Network (switches, iLO etc.)

Most of the infrastructure on my network uses static IP addresses but there is a range of IP addresses that are dynamically allocated to "transient" devices on the network, such as PC's, iPads, iPhones etc., for both home based devices and wireless "guest" devices brought in by visitors (provided that they have the access keys).

If a DHCP server was unavailable, then none of the transient devices would be able to get a new IP address when they tried to access the network for the first time or when their IP address lease had expired and the device tried to renew it. Rather than having the provision of IP addresses reliant on a single server, I wanted to provide a degree of redundancy for DHCP. Prior to Windows Server 2012, the windows server DHCP role did not support redundancy as such, but a work around was available by splitting the scope between two DHCP servers.

The two DHCP servers are both configured to provide IP addresses with the same range of addresses ("scope") but each server is configured to exclude 50% of the total range ("exclusion range"). Both servers are configured to have the same delay in issuing IP addresses in response to a DHCP request so it should be pretty much random which machine issues the IP address. In reality, it seems that the DHCP server with the lowest IP address tends to respond first, so about 75% of IP addresses tend to be allocated by the first DHCP server.

I configured split scope DHCP before my domain controllers were running Windows Server 2008 R2 so the configuration was done manually. Windows Server 2008 R2 includes a wizard that simplifies the process; right clicking on the "Scope" entry in the DHCP management console tree includes an "Advanced" menu option which exposes the "Split Scope" wizard.

If one DHCP server fails, the other server will continue to allocate IP addresses as required, provided that the scope is large enough to cater for the total umber of transient devices, i.e., total scope - exclusion range size, or 80 - 40 = 40 devices total.