Disaster recovery, as the introduction says, is the process for transitioning from one system to another system with as little interruption of service as possible. That's called a rollover, and there are three different ways of doing a rollover:
A hot rollover means that the infrastructure is completely mirrored at another site and that the backup site is always up and current with the primary site. This requires only a few adjustments to switch operations from the primary to the backup.
A warm rollover means that all of the elements for the backup site are in place (adequate network connections, all required applications and hardware) but the system is not actively running or necessarily configured. This can require some extra time to configure the machines and get the system running.
A cold rollover means that a site is available but there are few resources immediately available to set it up.
The obvious difference in the types of rollover is the time and expense necessary to set up the backup site. Hot and warm sites have higher initial expenditures to set up and run.
A mix of rollover types can be used, depending on the specific disaster scenario being planned. For example, a rollover plan for the loss of a single server could use a hot rollover easily and relatively cheaply by creating and keeping a virtual machine copy of the Directory Server instance which can be brought online within minutes. It would not even require keeping the virtual machine in a separate facility or network. On the other hand, a cold rollover could be planned for the loss of an entire data center or office.
Match the rollover process to the severity of the disaster scenario, your budget and available resources, and the likelihood of encountering problems.