Question: I can't afford to build a failover cluster. Are there any other options for providing fault tolerance?
Answer: Historically, the only way to make Microsoft Hyper-V virtual machines fault tolerant was to build a host server cluster. The problem with using failover clustering was that it required the use of a Cluster Shared Volume (CSV), which is expensive, to say the least. This requirement put Hyper-V clustering beyond the financial reach of many small and medium-sized businesses.
More on Hyper-V Replica
New VM replication tool for IT shops
Disaster recovery options for SMBs
Hyper-V Replica takes center stage at WPC 2011
Windows Server 2012 offers some lower cost alternatives to traditional Hyper-V clustering. One of the biggest changes that Microsoft has made in the new version is to remove the requirement for clusters to depend on a CSV. It is now possible to build a failover Hyper-V cluster without using shared storage. All you need is enough local storage in each cluster node to accommodate the VMs.
Another option is something called Microsoft Hyper-V Replica. Replication is similar to clustering, but it isn't a true clustering solution. When Hyper-V Replica is enabled, VMs (or rather the virtual hard drives that make up the VMs) are replicated from one Hyper-V host to another. The replica doesn't provide the same level of failover protection that a true cluster does, but you can manually perform both planned and unplanned failovers.
Another nice thing about the Hyper-V Replica feature is that you can create replicas using a minimum of two servers (a source server and a target server). Failover clusters require a minimum of three servers (either three cluster nodes or two cluster nodes and a file share witness). Being able to create a replica based on only two servers helps to reduce costs compared to failover clustering.
This was first published in December 2012