DHCP failover overview
In Windows Server® 2008 R2, there are two high availability options available for DHCP Server deployment. Each of these options is associated with some challenges.
- DHCP in a Windows failover cluster. This option places the DHCP server in a cluster with an additional server configured with the DHCP service that assumes the load if the primary DHCP server fails. The clustering deployment option uses a single shared storage. This makes the storage a single point of failure, and requires additional investment in redundancy for storage. In addition, clustering involves relatively complex setup and maintenance.
- Split scope DHCP. Split scope DHCP uses two independent DHCP servers that share responsibility for a scope. Typically 70% of the addresses in the scope are assigned to the primary server and the remaining 30% are assigned to the backup server. If clients cannot reach the primary server then they can get an IP configuration from the secondary server. Split scope deployment does not provide IP address continuity and is unusable in scenarios where the scope is already running at high utilization of address space, which is very common with Internet Protocol version 4 (IPv4).
DHCP failover in Windows Server 2012 enables administrators to deploy a highly resilient DHCP service to support a large enterprise without the challenges of the options discussed earlier. The main goals of the feature are the following:
- Provide DHCP service availability at all times on the enterprise network.
- If a DHCP server is no longer reachable, the DHCP client is able to extend the lease on its current IP address by contacting another DHCP server on the enterprise network.
The DHCP server failover feature provides the ability to have two DHCP servers provide IP addresses and option configuration to the same subnet or scope, providing for continuous availability of DHCP service to clients. The two DHCP servers replicate lease information between them, allowing one server to assume responsibility for servicing of clients for the entire subnet when the other server is unavailable. It is also possible to configure failover in a load-balancing configuration with client requests distributed between the two servers in a failover relationship.
DHCP failover in Windows Server 2012 provides support for a maximum of two DHCP servers, and the failover relationship is limited to IPv4 scopes and subnets. Network nodes using Internet Protocol version 6 (IPv6) typically determine their own IPv6 address using stateless IP auto configuration. In this mode, the DHCP server delivers only the DHCP option configuration, and the server does not maintain any lease state information. A high availability deployment for stateless DHCPv6 is possible by simply setting up two servers with identical option configuration. Even in a stateful DHCPv6 deployment, the scopes do not run under high address utilization, which makes split scope a viable solution for high availability.
DHCP failover architecture
Administrators can deploy DHCP servers running Windows Server 2012 as failover partners in either hot standby mode or load sharing mode.
In hot standby mode, two servers operate in a failover relationship where an active server is responsible for leasing IP addresses and configuration information to all clients in a scope or subnet. The secondary server assumes this responsibility if the primary server becomes unavailable. A server is primary or secondary in the context of a subnet. For instance, a server that has the role of a primary for a given subnet could be a secondary server for another subnet.
Hot standby mode of operation is best suited to deployments where a central office or data center server acts as a standby backup server to a server at a remote site, which is local to the DHCP clients (ex: hub and spoke deployment). In such deployments, it is undesirable to have a remote standby server service any clients unless the local DHCP server becomes unavailable. The figure below is an example of a hub and spoke deployment.
In a load sharing mode deployment, which is the default mode of operation, the two servers simultaneously serve IP addresses and options to clients on a given subnet. The client requests are load balanced and shared between the two servers.
The load sharing mode of operation is best suited to deployments where both servers in a failover relationship are located at the same physical site. Both servers respond to DHCP client requests based on the load distribution ratio configured by the administrator. See the following example.
Load sharing in a single site with multiple subnets
In the following example, two DHCP servers simultaneously provide IP addressing to clients on multiple subnets.
for more information and setting up an DHCP failover, see: http://technet.microsoft.com/en-us/library/hh831385.aspx