The Bottom Line
Identify the best options to provide disaster recovery for the various
services in your organization. When planning disaster recovery, application-
aware disaster recovery should be used first where possible, such as SQL Always
On, Exchange DAG, Active Directory multimaster replication, and so on. If no
application-aware replication and DR capability is available, another option is to
look at the replication capabilities of the SAN, such as synchronous replication.
Additionally, replicating at the virtual machine disk level, such as with Hyper-V
Replica, provides a replication solution that has no requirements on the guest
operating system or the application.
Master It Why is Azure Site Recovery useful?
Describe the types of failover for Hyper-V Replica. There are three types of
Hyper-V Replica failover. A test failover is performed on the replica server, and it
creates a clone of the replica virtual machine that is disconnected from the
network and allows testing of the failover process without any impact to the
ongoing protection of the primary workload as replication continues. A planned
failover is triggered on the primary Hyper-V host and stops the virtual machine,
ensures that any pending changes are replicated, starts the replica virtual machine,
and reverses the replication. An unplanned failover is triggered on the replica
Hyper-V host and is used when an unforeseen disaster occurs and the primary
datacenter is lost. This means that some loss of state may occur from the primary
virtual machine. When possible, a planned failover should always be used.
Master It In an unplanned failover, how much data could be lost?
Explain the automated options for Hyper-V Replica failover. Hyper-V
Replica has no automated failover capability. To automate the failover steps,
PowerShell could be used, System Center Orchestrator could be used, or, for a
complete solution, Azure Site Recovery could be used. The key point is that the
decision to use failover should not be automatic, because many conditions (such as
a break in network connectivity) could trigger a false failover. The automation
required should be the orchestration of the failover after a manual action is taken
to decide whether a failover should occur.