Leveraging virtualization for data protection
Consolidation of physical servers through virtualization is the norm today, and no wonder. Virtualization can deliver better hardware utilization, improved business and IT agility, and lower costs through efficiencies in power, cooling and use of data center space.
Yet consolidating many physical servers into a few raises concerns about data protection and business continuity. How will virtualization impact data backup and recovery, and how can you prevent data loss and downtime of virtual applications and servers? In fact, virtualization projects are an excellent opportunity to improve data protection by evaluating your current processes, and enhancing them via the unique characteristics of virtualization technology.
Define IT objectives
"The first thing is to analyze your recovery time objective (RTO) — how quickly you have to have the application back and running — and your recovery point objective (RPO) — how much data you can afford to lose," says Greg Schulz, founder and senior advisor at the Server and StorageIO Group, and author of Cloud and Virtual Data Storage Networking1. "You can do this for your whole environment or for individual virtualized applications. Then you can evaluate the technologies and tools that will address these requirements."
Before virtualization, many companies run a file-based backup agent on each physical server that streams data to a backup server and then, most often, to tape. After virtualization, this same process could be problematic if the many backups running on the physical server cause I/O bottlenecks that slow the process or production applications. Tape also can limit the speed of backup and recovery, as well as the ability to deduplicate data. And file-based backup alone does not facilitate business continuity should an entire virtual machine or the physical server fail.
Use next-generation backup
There is a better way: take advantage of next-generation backup and recovery solutions intentionally designed for virtualized environments. Typically these leverage disk-based backup built around a centralized network storage architecture and managed by virtualization-aware software. Best practices for network-efficient backup can include online recovery with integrated SAN tools, automated off-host backup to disk appliances and deduplication at both source and target locations.
As an example of one promising approach, consider several production servers running many virtual machines alongside a backup server, all attached to a storage area network (SAN). "Now it's possible for the backup server to talk directly to the underlying storage, streamlining the process by offloading the job from the servers running the virtual machines," says Schulz. "That shows how rearchitecting data protection in a virtual environment can improve performance."
Snapshots speed recovery
Another plus is the ability to improve business continuity through SAN-based snapshots of virtual machines. "Virtual machine snapshots are superior to traditional backups for tasks such as service-pack updates, hotfix installs and application updates, because they can be performed in seconds and recovered in about the time it takes to reboot the server," says David Culbertson, director of technology at Solve IT, a solution provider in Lakewood, Colo. "Short-term backups can also be accomplished at the SAN level by snapshotting the entire volume housing the virtual machines. This is a convenient way of protecting the entire server against a failure," he says. In addition, replicating snapshots off-site can enable business continuity.
Of course, primary storage still requires protection, which could be to tape, disk or the cloud. Backup to a disk appliance can provide network-efficient backups. Features to look for include single-console management, user self-service for granular backup and file recovery, and automated compression, encryption and archiving for offline storage. Backup and recovery times can be reduced through data deduplication.
These are just a few possibilities for leveraging virtualization to improve backup and recovery. As Schulz says, "Virtualization is an excellent opportunity to rethink, rearchitect and modernize how you're protecting your data and ensuring business continuity."
Better virtual backups
Consider the trade-offs: When making changes to backup processes, be sure to consider RTO and RPO, and the trade-offs involved. The more aggressive the targets, the more costly the solution.
Evaluate server utilization: The ability of a server to power multiple virtual backups depends on its utilization. Highly utilized servers will benefit from backup from shared storage, or from proxy servers accessing virtual snapshots.
Pick the right target: Virtual backups can be sent to tape, disk or the cloud, each with a unique profile for convenience, performance and cost. Many companies find disk represents the best value.
Chose vendors widely: The best products for virtual data protection come from vendors that partner closely with virtualization providers such as Citrix, Microsoft and VMware. A key criterion is the depth of support for your hypervisor of choice.
1 Cloud and Virtual Data Storage Networking, CRC Press, August 2011.
About the author: Tom Farre, a freelance journalist and the former editor of VARBusiness, has been covering information technology for more than 20 years.