Disaster recovery can mean vastly different things to different organizations. Every business will have different recovery time objectives and recovery point objectives. But they all face the challenge of designing an effective and cost-efficient recovery plan.
Over the past several years, virtualization has changed the game. Built-in features from many virtualization vendors often provide at least a basic level of protection, but more substantial disaster recovery (DR) costs big bucks. So, how do you balance disaster recovery costs with the value of your workloads and data? This week, our Advisory Board discusses how to approach disaster recovery spending in a virtualized environment.
Jason Helmick, Interface Technical Training
When I started in IT, the official disaster recovery plan at many companies was a single-page document that contained one thing: the phone number to IBM's support services. The disaster recovery costs were whatever IBM charged.
DR planning is much more complex today, and if you're an IT pro who knows the meaning of business continuity, recovery time objectives (RTOs) and recovery point objectives (RPOs), then you're already in the deep end. For the IT pro in a smaller company that relies on service continuity while facing a constrained budget and doesn't have the planning resources of a big company, you have the same cost-cutting weapon as your peers at larger companies: virtualization.
In the "old school" days, we looked at separating mission-critical services from normal services. Mission-critical services -- think authentication and line-of-business applications -- needed to be rapidly resilient. Losing one of these services cost the business money -- we measured nonrecoverable funds per hour back in those days. Organizations often implemented a preventative, redundant technology, such as clustering, if the cost of downtime exceeded the expense in implementing the redundant solution. Normal services that would have a lower cost to the business if they were unavailable, such as email, were often relegated to a simple backup-and-restore process because funds had been allocated for the mission-critical services. The trouble is, clustering is expensive and means specialized hardware, implementation and maintenance costs. Even two node clusters can easily be beyond the budget of a small company.
Virtualization has reduced the expense of implementing rapid resilient recovery, mainly by removing the specialized hardware expense and dramatically reducing the implementation and management expense. With virtualization platforms from Microsoft and VMware, IT pros can choose among built-in operating system or application recovery options, or one of the many virtualization supplied options. Now, the normal services once left to long recovery time tactics can now have ready-to-boot virtual copies for near real-time recovery. By using virtualization, even small companies can afford to increase reliability and coverage across all their services.
Explore all your recovery options from your preferred virtualization vendor. Understand how to use them to provide resilience to your critical and noncritical services. Regardless of your company's size, I bet the next disaster recovery meeting will revolve around these cost-reducing options.
Robert McShinsky, Dartmouth Hitchcock Medical Center
Restoration used to mean tape backups to physical servers. This is a fine disaster recovery method, albeit with the possibility of data loss. Server virtualization added much easier restoration capabilities with both whole systems or file-level recoveries. There was still some data-loss potential, but it was a fairly cost-effective method and provided quicker restoration, since a replacement virtual host system was all that was necessary. Today, there are even more disaster recovery options that leverage virtualization. Some fit particular sizes of organizations, such as Hyper-V Replica for small and medium-sized businesses (SMBs); or solutions for larger environments, such as VMware's SRM or true storage area network replication with GeoClustered hosts. The cloud poses another option that is maturing with more capabilities, which let you directly provision or replicate VMs to the cloud. All of these options benefit from virtualization, but each comes with its own price tag. In-box solutions often cost less, but traditionally have less scalability. With scalability comes more complexity and cost, but greater capabilities.
From my experience, as frustrating as it might be, the decision will come down to upper management to identify whether business continuity is based on minutes, hours, days or weeks. Informing them of accurate times and hurdles to recover the business in the event of minor or major disasters will be the key.
Dave Sobel, Level Platforms
Effective disaster recovery plans are tough. It's easy to spend too much or too little, and missing can be disastrous. So how do you plan? Measuring the cost of downtime is critical to making the right decision. How much does being offline cost, both in repair costs as well as lost revenue? Understanding this formula on a per-workload basis is the first step, as different workloads have different values. Virtualization has made disaster recovery costs go down significantly and offered many new options for recovery that didn't exist when the software and hardware were bound together. Where previously a restore would be a specialized effort in the case of a disaster, there are many new ways to approach the problem. In fact, many offer significant savings by leveraging virtualization and cloud-based technologies (which often have virtualization at their core). This has moved sophisticated disaster recovery costs down to the point where SMBs can afford it.