Clustering

Clustering in a virtual environment can be of two main types: parent clustering and child clustering.

Parent clustering uses two physical computers, and both host operating systems must run the Failover Clustering service.

The clustered child virtual machines run on the parent computer, either in the active-passive mode with one node on standby or in the active-active mode with both nodes running applications.

If a parent node fails, the Failover Clustering service fails over all of the virtual machines from the failed node to the healthy node.

In child clustering, virtual machines run either on one or more host computers.

The difference is that in child clustering, each child operating system runs the Failover Clustering service and monitors the clustered applications or services locally.

If a clustered application or service fails, the Failover Clustering service fails over the application or service to the other child node in the same way as hardware-based clustering.

Parent clustering has the advantage that the virtualized applications, network services, or operating systems do not have to be cluster-aware, because the failover process occurs at the virtual machine level.

However, with child clustering, the child operating system must support Failover Clustering.

You can combine parent and child clustering to provide seamless transfer of network services, applications, and virtual images between virtual machines and between physical computer nodes.

For both parent and child clustering, iSCSI-based storage provides the best storage solution.

With iSCSI, you can remove the two-node restrictions from older technologies, such as parallel SCSI connections, and use eight-way clustering. iSCSI also matches multiple virtual machines to different storage locations.