virtual workloads. Beyond that, though, is making services available in unplanned
events such as power outages, host crashes, and natural disasters. Windows Server
2012 greatly improved Failover Clustering, which is the backbone of Hyper-V high
availability. However, what many customers asked for was a disaster-recovery (DR)
feature that would allow an asynchronous replication of virtual machines from one
datacenter to another. Hyper-V Replica provides this capability exactly, allowing the
virtualized storage of a virtual machine to be replicated to a DR location Hyper-V
server every 5 minutes in addition to providing numerous failover options, including
the ability to test failover without impacting production replication. I cover high
availability and disaster recovery in great detail later in the book, and I don’t consider
Hyper-V Replica the answer to all DR situations. Hyper-V Replica, which provides
asynchronous replication between a primary VM and a replica VM, is one available
tool that works well in specific scenarios.
WHY IS ASYNCHRONOUS REPLICATION A GOOD THING FOR
DISASTER RECOVERY?
Typically, synchronous is best for any kind of replication. With synchronous
replication, a change made to the primary store is not committed until it is also
written to the secondary store. For the best assurance of data integrity and to
ensure no loss, this is a good thing. However, synchronous replication has a
substantial cost. The connectivity required for synchronous replication needs to
be resilient and fast enough, with a low enough latency to ensure that the
performance of the primary workload is not negatively affected. For the
replication of a virtual machine across datacenters, only the highest levels of
connectivity would enable the storage replication without affecting the primary
workload. Although these solutions are possible, they are typically part of SAN
solutions, which are usually costly. With asynchronous replication, the primary
workload is not affected, and the changes are replicated to the secondary store as
quickly as possible or on a fixed interval. This achieves a good level of protection
without requiring very fast, low-latency network connections, but it is not real-
time replication. In the event of an unplanned failover to the DR site, a few
minutes of data may be lost, but in a true disaster, a few minutes of state loss is
typically accepted. Asynchronous brings disaster recovery to all workloads rather
than just the tier 1 services that can utilize SAN-level synchronous replication.OTHER CAPABILITIES
Windows Server 2012 Hyper-V introduced many other capabilities that greatly change
virtual environments:
Virtual Fibre Channel support that allows virtual machines to communicate
directly with Fibre Channel–connected SANs, which is a necessity for guest
clustering scenarios that need shared storage and cannot use iSCSI