Many organizations write a DR plan merely to satisfy auditors. When disaster strikes, the plan is set aside and recovery becomes an ad-hoc exercise. But new virtualization technologies have made it more feasible, efficient and affordable to implement a DR plan.
In recent years, business and regulatory requirements have driven the need to have a comprehensive disaster recovery (DR) plan. Unfortunately, this trend has led to a rush of good-enough-for-the-audit solutions that few people expect to actually work. Essentially, a DR plan has become just another box that you need to check.
The truth is, there are a number of factors that have plagued disaster recovery efforts over the years, resulting in a disconnect between the creation and execution of a plan. A major issue is the nature of migrating workloads from one physical machine to another. A backup of a Windows Server 2008 virtual machine (VM) on Server A is only useful on Server B if the hardware is nearly identical. That means every time you buy a server, you need a matching one for your DR plan, and that isn’t very cost effective.
The workaround usually involves manually reinstalling the operating system and applications on the DR server and pulling the application data from the backups. This process is time consuming and prone to errors, which is the very definition of inefficiency. But don’t worry, for there is a better way: a virtual DR plan.
Virtual DR plans to the rescue
Virtualization has broken the chains that link an OS to a specific piece of hardware. The OS is encapsulated and highly portable, eliminating one of the biggest barriers to an efficient and actionable DR plan.
I’ve seen the difference firsthand, when I replaced a physical DR plan that failed to complete a test in 72 hours. I switched to a virtual DR plan, and it was executed and verified in less than 24 hours. Even better, the virtual DR plan was less expensive!
Since that experience, I was sold on the benefits of a virtual DR plan. Even if your production workloads aren’t run in VMs, you can use a variety of physical-to-virtual conversion tools to create virtual copies of physical servers. This ability opens the doors for a 100% virtual DR plan, regardless of how much of your production workloads are virtualized.
Getting over that recovery hump
Even with an efficient virtual DR plan in place, many projects get hung up on data recovery. Tapes are slow, and replicating storage can be complicated and expensive. For a low recovery time objective (RTO) and recovery point objective (RPO), storage replication is king.
A limiting factor, however, is that many storage replication products require matching storage arrays or expensive replication devices. Again, if you need a low RTO and RPO, you may not have an option. But if you can accept a slightly higher RPO, there are new replication products that can bridge the recovery gap.
Both VMware Site Recovery Manager 5 and Zerto Virtual Replication can replicate storage between two VMware infrastructures that don’t have matching storage arrays or replication devices. In fact, these products can replicate local disks and don’t even require a storage array.
The RPO of these products is in the 15-minute range, which may be tolerable for most organizations, but it’s not quite in the synchronous or near-synchronous range that some financial organizations need for portions of their infrastructure. Still, software-based replication products provide a high degree of simplicity and flexibility at a reasonable price.
Thanks to virtualization, disaster recovery options are more mature than ever. With these tools, there really isn’t an excuse for avoiding a comprehensive virtual DR plan that can be easily and accurately tested.