The biggest problem being faced by organizations transitioning to the Cloud is disaster recovery of their virtual machines especially in instances where virtualization itself is used in disaster recovery plans as an option.  As a result, a number of approaches to disaster recovery planning have been formulated, facilitated and executed by VM developers.  Let us look at the Virtual Machine disaster recovery features in context, to understand the direction of thinking in VM disaster recovery.

In the event of disaster, the Virtual Machine must failover to a VM in a remote location. The vCenter Site Recovery Manager is software, built into VMware, to automate the failover process. Where geo-clustering architectures are used, the cluster can recover more than the virtual machines. There are a number of other alternate packages that are cheaper, less automated and support failover with varying levels of sophistication.

Data administrators have to perform several tasks before the Virtual Machine disaster recovery gains shape. As a first step, the onsite data administrator must establish data store replication procedures. This may be between the local site and the remote site or between two local machines (some VM DR requires that the two machines have different IP addresses) or two remote machines. The replicated data stores must be identified, checked for accuracy and configured with the primary site for failover procedures. This involves selection of the virtual machines and remapping of the hardware / software, putting in place automated monitoring protocols and the creation of a disaster recovery plan.

It is possible to define multiple VM disaster recovery plans for failover and redundancy. Data administrators may select the most appropriate plan for a given situation with varying failover capabilities.  For instance, the full VM recovery plan may be different from a partial recovery plan.  The plan may or may not include manual steps.

VM Disaster recovery testing must be supported by the DR Plan. The test process must facilitate both, complete failure testing and partial failure testing. This has to be typically a scheduled activity. The failover has to be orchestrated and can be very time consuming. The primary site may take time to be brought back online after the failover testing is initiated. The recovery to the primary site may also involve re-establishing data store replication, re-identification of protected data stores, re-selection of protected VMs, re-mapping of all site inventories, and re-creation of the failback recovery plan.

Before we end this discussion, it is important to understand that not all VM DR software support all features required for effective data recovery.  Users may have to use combinations of different software to come up with a DR plan that works for them.

Securstore is here to help you with your Virtual Machine Disaster Recovery needs. Contact us any time.