When Disaster Strikes: A Virtual World Speeds Real Recovery

At most business sites, disaster recovery (DR) planning is treated as only an extension of IT data backup. That datacenter myopia fosters dangerous underestimates of the potential scope and cost of a real disaster. The goal of any IT disaster recovery plan is to protect the most business-critical processes and minimize unplanned downtime.

"A significant infusion of resource virtualization will enhance the ability of IT to restore processing services while simultaneously making the plan more cost-effective with the help of the advanced resource management capabilities of the IBM Systems Director family"

Selecting the optimum disaster recovery solution will come down to balancing a solution's cost with the potential financial loss from a disaster. Along those lines, a significant infusion of resource virtualization will enhance the ability of IT to restore processing services while simultaneously making the plan more cost-effective with the help of the advanced resource management capabilities of the IBM Systems Director family. To get a sense of how important a well-optimized DR plan can be, just consider what happened in Buffalo, N.Y. this past October.

A Cold Wind in Buffalo
On Thursday morning Oct 12th, Buffalo woke to a fall day that looked ideal for viewing the brilliantly colored fall foliage. Before the day was over, a freak lake-effect storm from Lake Erie would dump a record 8.6 inches of heavy wet snow, with an additional 12 inches of snow falling the next day. By the time it was over, Buffalo lay paralyzed as tree limbs, weighed down by leaves and heavy wet snow began to snap and take down electric power lines blacking out over 300,000 homes and businesses. It would take 10 days for the entire region to crawl out from under the disastrous storm.

The Buffalo blizzard is just one example of the kind of external event that can affect business continuity. With employees stranded and no electric power, companies of all sizes in the Buffalo area were confronted with a business continuity problem that had nothing to do with hardware failure or software errors.

To deal successfully with such a disruptive event, a business needs to identify its critical business processes and the costs of their downtime. These costs include salaries paid to staff unable to undertake billable work, lost ability to respond to contract opportunities, the cost of interest on lost cash flow, and potential penalties from failure to provide tax or compliance reports. Once the key business services and the risks to those services have been established, an IT disaster recovery plan, which details the time parameters to restore full business services and acceptable service levels during the recovery-period, needs to be put in place.

Disaster Recovery Is a Matter of Time
An IT disaster recovery plan is all about time. It starts with the time it will take to make data available within a functional IT infrastructure. Next is the time it will take to restore IT operational processes. Finally, there is the time it will take to restore business processes. As a result, recovery time becomes the critical defining characteristic for a disaster recovery plan, and the major costs associated with that plan are the costs of complying with those time constraints.

Storage virtualization plays a key role in cutting the recovery time and controlling disaster recovery costs. To minimize the time it takes to make data available after a disaster, IT can implement a number of data replication strategies including point-in-time data copies and continuous data mirroring. Without virtualization, the physical storage infrastructure for the original data and the copies will need to be identical.

The need to implement an identical storage configuration when managing physical devices negates one of the key elements in cost control: the ability to define a performance level that is acceptable during the recovery transition period. It also opens the door to inefficiencies and extra costs associated with over-provisioning. Given that the recovery transition period is by definition a finite transition period, less capacity, higher latency, and lower throughput may be entirely acceptable for the resources supporting the duplicate data in order to reduce costs. With storage virtualization, the constraint of duplicate physical storage no longer negates the savings potential.

The introduction of block-level storage virtualization for all hosts and applications provides host systems and applications with a logically consistent view of disk storage resources that can be mapped to the physical storage. With the hosts systems dealing with a consistent logical representation, IT is free to optimize the supporting physical storage infrastructure at will.

More importantly, IT automatically derives the benefits of virtualized replication services to simplify and lower data protection costs. By virtualizing replication services, IT operations managers have a common method for performing all local and remote data replication. They also have the freedom to choose high-end storage for the primary data and low cost storage for the replicated data to be used in a disaster recovery transition. The IBM® SAN Volume Controller (SVC) fits perfectly into this scenario as a means to improve business responsiveness and continuity.

Real Fast Recovery
The second issue in the disaster recovery process is the time it will take to restore IT operational processes. At this point, system virtualization can play an equal role with storage virtualization as a means to create efficiencies and save costs. Most system failover scenarios also rely on duplicate resources. Once again, however, the needs of the transition period may not be as great as the needs of full production. Virtualizing systems is a way to readily deal with that issue. What's more, the new Virtualization Manager Extension to IBM Director is an ideal way to provision and manage those systems in a disaster recovery transition.

By running systems as virtual machines, IT operations managers can readily move virtual servers from one physical server to another. Using the Virtualization Manager Extension to IBM Director and exploiting IBM Director's deep hardware knowledge of the physical server, predictive data on the health of the physical server can be gathered and policies put into place that trigger the migration of virtual machines off of systems that may fail. What's more, in a remote hot-site scenario, applications that run on Linux virtual machines with no processor restrictions, can be consolidated on any server, whether IBM System z™, System p™, System i™, or System x™.

The bottom line for disaster recovery is that it is a matter of when not if. Whether the problem comes about through a fast-spreading Internet Warhol Worm, a natural disaster, or an external event such as a major accident or a labor strike, a disaster will happen and business continuity will be impacted. To deal effectively with the problem and not risk new business, run up unnecessary costs, and worse, diminish brand equity, a business continuity plan needs to be in place that balances true risks with real costs. To tilt that balance in favor of higher continuity response, system and storage resource virtualization can lower contingency costs, in some cases dramatically. Along those lines, the IBM Systems Director family of virtualization solutions provides the power to reinstate order from the chaos.