1. Previous approaches to implementing DHCP availability
Traditionally, DHCP server availability has been implemented on
Windows Server–based networks using one or more of the following
methods:
-
Split scopes This approach
involves splitting the IP address pool of a scope between two DHCP
servers, typically by assigning the primary server 80 percent of
the addresses in the scope and the secondary server the remaining
20 percent of the addresses. That way, if the primary server goes
offline for any reason, DHCP clients on the subnet can still
respond to lease renewal requests from the secondary
server.
-
Server cluster This approach
involves using the Failover Clustering feature of Windows Server
2008 or Windows Server 2008 R2 to cluster DHCP servers so that if
the primary DHCP server in a cluster fails, the secondary server
can take up the slack and continue leasing addresses to
clients.
-
Standby server This approach
uses a hot standby DHCP server with scopes and options configured
identically to your production DHCP server.
Each of the preceding approaches has the following
disadvantages, which make them of limited usefulness in ensuring DHCP
server availability:
-
The split-scope approach provides limited IP availability
during outages. As a result, some clients might not receive
addresses during a long-term DHCP server outage. In addition, if
your DHCP server scope is currently running at high
utilization—which is common for Internet Protocol version 4 (IPv4)
networks—splitting the scope might not be feasible.
-
The DHCP server-cluster approach has only one DHCP database
located on the cluster shared storage. That means there is a
single point of failure for DHCP services on your network. In
addition, implementing Failover Clustering requires relatively
complex setup processes and maintenance tasks.
-
The hot-standby approach requires both careful configuration
of the standby DHCP server and manual intervention on the part of
the administrator to ensure the failover transition when your
production DHCP server fails or goes offline. There is also
additional complexity in this approach when DHCP is configured to
automatically update DNS records, as is recommended in an Active
Directory environment.
Note
Spanning multiple subnets
Routers usually block DHCP packets from being forwarded from
one subnet to another because DHCP packets are broadcast traffic. If
an organization’s network consists of multiple subnets, using DHCP
for dynamic address assignment requires either of the
following:
Understanding DHCP failover
DHCP failover is a new approach to ensuring DHCP availability
that is included in Windows Server 2012. With this approach, two DHCP
servers can be configured to provide leases from the same pool of
addresses. The two servers then replicate lease information between
them, which enables one server to assume responsibility for providing
leases to all clients on the subnet when the other server is
unavailable. The result of implementing this approach is to ensure
DHCP service availability at all times, which is a key requirement for
enterprise networks.
The current implementation of DHCP failover in Windows Server
2012 has the following limitations:
DHCP server failover can be implemented in two different
configurations:
-
Load-sharing mode Leases are
issued from both servers equally, which ensures availability and
provides load balancing for your DHCP services. (This is the
default DHCP server failover configuration.)
-
Hot-standby mode Leases are
issued from the primary server until it fails, whereupon the lease
data is automatically replicated to the secondary server, which
assumes the load.
A typical scenario for implementing the load-sharing approach
is when you want to have two DHCP servers at the same physical site.
If the site has only a single subnet, all you need to do is enable
DHCP failover in its default configuration. If there are multiple
subnets, deploy both DHCP servers in the same subnet, configure your
routers as DHCP relay agents (or deploy additional DHCP relay agents
in subnets), and enable DHCP server failover in its default
configuration.
When implementing the hot-standby mode approach, you can
configure a DHCP server so that it acts as the primary server for
one subnet and as the secondary server for other subnets. One
scenario where this approach might be implemented is in
organizations that have a central hub site (typically, the data
center at the head office) connected via wide area network (WAN)
links to multiple remote branch-office sites. Figure 1 shows an
example of an organization that has DHCP servers deployed at each
branch office and at the head office. Branch-office servers are
configured to lease addresses to clients at their branch offices,
while the central server leases addresses to clients at the head
office. Each branch-office server has a failover relationship with
the central server, with the branch office assuming the role as
primary and the central server as secondary. That way, if a DHCP
server fails at a branch office, the central server can take up the
slack for the remote site. For example, the DHCP server at Branch
Office A is the primary server for the scope 10.10.0.0/16, while the
DHCP server at the Head Office is the secondary for that
scope.
