Architecting the Right Disaster-Recovery Solution
After reading this chapter, you may think that you will just use Hyper-V Replica for all
of your disaster-recovery needs. The reality is that should not be the case. It’s a good
technology, but better options may be available for specific workloads and
applications.
Your first preference should be to use an application’s replication and disaster-
recovery capabilities if it has them, because it’s always better for the application to be
aware of a failover and manage its own data if possible. For example, SQL has its
Always On replication and failover technology, which means that if I had data in SQL
Server, I would always use Always On first. Likewise, Exchange has database
availability groups and replicates mailboxes. In that case, I would use that technology.
Active Directory has multimaster replication, which means that I would simply have
domain controllers running in my DR location that would replicate from my primary
location for normal operation.
If the application is not stateful, such as a website, then I could have instances of the
application running in the primary and DR locations and use network load balancing
to spread the load. Another option is to update DNS to point to DR instead of primary.
Although there are some considerations around caching of DNS records, solutions
exist.
If the storage has some kind of synchronous replication capability that could be used
as if it were a single logical piece of storage that would allow a cluster to be used, my
next preference would be to treat the replicated SAN as shared storage and enable any
cluster-aware application to leverage the SAN. This could include virtual machines
running on the replicated SAN. This could also include Storage Replica’s synchronous
option that integrates with clustering.
If none of those other options were available, I would use Hyper-V Replica. That is not
to say that it’s not a great solution, but a hypervisor-level asynchronous replication
solution that is not application aware is simply not as rich in its functionality as one
that is part of an application. Most organizations have a lot of applications that don’t
have their own replication technology and can’t run on a replication SAN, so a huge
number of virtual workloads can benefit greatly from Hyper-V Replica.