When formulating a backup strategy, it's important to identify your needs and then match each need to the most appropriate storage product. In many cases, a backup strategy will involve numerous disk-based technologies, and each will offer a price/performance tradeoff that must also be considered. Every backup vendor sees the marketplace differently and offers their own unique take on the practices and principles needed to formulate a solid strategy.
Recoverability and availability
The role of backups has changed in the last few years. Traditional backups simply ensured that a copy of corporate data was available just in case "something" happened. Today, the cost of downtime and criticality of corporate data has shifted the emphasis to recovery. Administrators expect files to be lost or damaged, so the real objective now is to have a timely backup available and guarantee that the backup can be recovered quickly enough to minimize impact on the business. "No one ever lost their job for missing a backup window," says George Symons, chief technology officer of the information management software group at EMC. "On the other hand, not being able to recover your Exchange server could cost you your job."
Symons sees the emerging role of disk technology in backup strategies but warns against treating all data the same way. "You have different requirements for different applications," Symons says. "To try backing everything up the same way and recover it the same way is what doesn't work." Achieving high performance across the board becomes too expensive, but a single inexpensive backup system may not offer the recovery time that is needed. Symons suggests matching the backup platform, such as a virtual tape library (VTL) and continuous data protection (CDP) with the recovery needs.
Many organizations also do a poor job when it comes to testing -- an oversight that can have disastrous results when recovery is needed. "One of the biggest problems people have had over the years is that they don't verify and test what they can recover," Symons says. Any backup strategy should include specific provisions for verification and periodic recovery testing. Testing keeps recovery skills sharp, helping IT personnel perform better when trouble strikes.
Finally, Symons points out that backups and archives do not serve the same purpose and should not be used interchangeably in the enterprise. Archives should retain data that doesn't change but must be accessible for some prescribed period of time (e.g., to meet compliance needs). Conversely, backups are not intended to be readily accessible (e.g., to meet compliance audits). For example, companies that attempt to use backups for legal discovery can spend up to $1 million or more on a single discovery request when working with backups. By matching backup and archival storage needs, a company can better serve both needs and actually reduce backup requirements by offloading archival data first.
Don't ignore the remote office
Backup strategies often include careful consideration of key business applications and retention/compliance requirements, but remote offices are sometimes overlooked as a critical data repository. This is particularly true in larger global enterprises where 24/7 operation makes it difficult to accommodate any backup window. Regional sales offices and bank branch offices are just a few examples where vital corporate data may need to be accounted for in a backup strategy. "The amount of data that is actually critical and that sits on the edge [of the data center in remote locations] has become much larger," says Manish Goel, vice president and general manager of data protection and retention solutions at Network Appliance Inc. (NetApp). As an example, Goel notes that it may be critical to backup the e-mail system in a branch office where the idea may have been impractical, even unnecessary, just a decade ago. This places a new emphasis on remote backups in any strategy.
Goel strongly advises against scaling tape in today's backup environment and echoes the growing importance of disk in three key backup strategies that appear to work well. First is the trend away from conventional tape in favor of VTL. "It's giving you the immediate benefit of disk but in a way that is completely non-disruptive to the existing backup methodology," he says.
The second trend is direct disk-to-disk backup that simply implements an exact copy of the primary storage. Although disk-to-disk backups use a slightly different process than VTL, the exact disk copy can be used for other business purposes -- including offloading the disk contents to tape. "It's basically a disk-to-disk-to-tape strategy," Goel says. "And what we find is that they run in that mode for anywhere between 6-12 months. They end up over time eliminating tape from their environment."
A third important trend is remote office consolidation strategy where backup tasks are eliminated from remote office, and WAN links are used to facilitate backups to a central location. Goel points to one customer in the energy industry with over 1,000 global remote offices to contend with. By implementing NetApp SnapVault appliances in remote offices, those remote offices are streaming their backup data to about 10 regional data centers. "They now feel 100% comfortable that their remote offices are protected, and the data is actually being backed up," Goel says. "More important, their total cost to do that has gone down by 65%."
Backups as a dynamic resource
With data volumes growing, the time available to establish a restore point (a.k.a. "backup") is dwindling. In some cases, backup times are actually pushing beyond the evening or weekend and extending into the workweek. This generally means that fewer backups are created, leaving more data vulnerable to loss in the event of a disk fault, virus or other problem. Vendors are utilizing disk storage systems for faster backups, which also allow more frequent backups -- offering users far more granularity in their recovery point. Enterprises that demand more recovery points are often incorporating snapshot and CDP technology into their backup strategies.
Rather than make a full weekly backup or nightly incremental, snapshots can capture the system state several times each day, even several times each hour. CDP is far more granular and can track a system state right down to the last disk write -- resulting in a recovery point of almost zero. "The process of backing up effectively goes away," says Kirby Wadsworth, senior vice president of marketing and business development for Revivio Inc., "We have a continuous flow of information from the primary [production system] to the secondary [backup storage] in real time."
Wadsworth notes that CDP data on disk can be utilized in ways that are simply impossible with tape. For example, the data contained in a CDP instance can be loaded into a business intelligence application or copied to a lab server where new applications or patches can be installed and tested prior to a rollout. A CDP instance can also be archived as part of the backup strategy. "People mount those new disks [the CDP instance] on a backup server and then back them up to either a VTL or tape," Wadsworth says.
Users considering CDP technology will need to make some fundamental changes to the way they view or use backups. Rather than treat backups as a library of storage events, users must approach CDP-based backups as a dynamic resource that is always changing. "You need to move away from making, managing and storing copies," Wadsworth says. "Instead, at the push of a button you can get any copy thing from any time -- that is a huge philosophical and strategic change in mindset." The hardest part of this change is relating events to the overall CDP timeline. For example, CDP can help to overcome virus infiltration, but users must be able to identify the point in time where the virus first struck.
Go to the next part of this article: Backup strategies: User perspectives
Or skip to the section of interest: