cannot enable Hyper-V Replica for virtual machines that leverage virtual Fibre
Channel, iSCSI, or pass-through storage. The virtual machine must use virtual hard
disks for all storage because it is through the VHD implementation that replication is
enabled. If you use virtual Fibre Channel or iSCSI, the assumption would be that the
SAN is performing some level of replication between the primary and replica
locations, which means that it would make more sense for that SAN replication to
replicate the VHDs of the virtual machine as well as the LUNs attached to the VM
using virtual Fibre Channel/iSCSI. You would not want two replication technologies
used that would be out of sync with each other.
Because Hyper-V Replica uses asynchronous replication, there is a period where the
replica is missing some of the data from the primary source. Potentially, in an
unplanned failure, a certain amount of data may be lost. The exact amount of loss
depends on how frequently the replica is updated. In Windows Server 2012, this was
every 5 minutes, but in Windows Server 2012 R2 and above, it can be every 30
seconds, every 5 minutes, or every 15 minutes. This possible data loss needs to be
compared against the RPO of the application. If the application has an RPO of
5 minutes, you can replicate at a 5-minute or 30-second interval. If the RPO is 1
minute, you must replicate at the 30-second interval and also ensure that there is
sufficient bandwidth to handle the transmission of logs from the source to the replica.
The good news is that because the replication is asynchronous, the introduction of the
replica does not lead to any performance degradation on the source virtual machine
and does not require very fast, low-latency network connections between the source
host and the replica host.
The use of asynchronous replication by Hyper-V Replica opens up the use of
replication for disaster-recovery scenarios to many more scenarios and types of
companies. These are some key ones that are often considered:
Datacenter-to-datacenter replication for tier 1 applications for organizations
without SAN-level replication such as small and medium-sized organizations
Datacenter-to-datacenter replication for tier 2 applications for organizations with
SAN-level replication but that don’t want to use the SAN-level replication for
applications that are not tier 1
Replication from branch office to head office to protect applications hosted at the
branch location
Replication from hoster location 1 to hoster location 2 for hosting companiesSmall organizations that do not have a second datacenter can replicate to a hoster as
the secondary datacenter for DR needs or even to a consulting organization’s
datacenter for their clients.
Many more scenarios exist, including anything that is enabled through the
asynchronous replication of a virtual machine. The key point is that with Hyper-V
Replica, the ability to replicate virtual machines is now an option for any organization
with two locations. As I cover later in the chapter, it’s an option even if an