IT administrators are always looking for ways to revamp their backup strategies. Today's exciting technologies and capabilities provide an added incentive to take a fresh look at backup procedures. To identify and take advantage of these advancements, consider these recommendations for making the most of these technologies and for devising a sound Windows backup strategy.
During the past few years, two key developments have changed the landscape for performing backups of Windows and Windows applications such as Microsoft Exchange Server and Microsoft SQL Server. The first development is new hardware capabilities. We've come a long way from the 8mm helical scan tape devices I used for server backups 10 to 15 years ago. Today's new technologies, such as AIT, Linear Tape-Open (LTO), and SuperDLT (SDLT), have capacities and transfer speeds of a few hundred gigabytes per tape—equal to the performance and capacities of the hard disks we used 10 years ago. But the advances in high-speed and high-capacity backup facilities haven't kept up with our voracious storage appetites. On the hardware side, advances in Network Attached Storage (NAS) and Storage Area Network (SAN) technologies have provided new storage options in the form of hardware-based snapshots and clones and data-replication technologies such as the HP StorageWorks Data Replication Manager (DRM) and EMC's Symmetrix Remote Data Facility (SRDF). But without an OS- and application-supporting framework on which to build these technologies, none of the new hardware advances can handle the Windows backup challenge.
The second development is new Windows technologies such as Microsoft Virtual Disk Service (VDS) and Microsoft Volume Shadow Copy Service (VSS), which let Windows and Windows applications leverage hardware capabilities natively. Both VDS and VSS let third-party vendors support Windows-based, state-of-the-art backup-and-recovery and storage-management technologies. Now open and Microsoft-supported solutions are available that both vendors and customers endorse.
These two developments can help you get a better handle on backups (and ultimately provide improved recovery as well). Of course, you need Windows Server 2003 and new hardware to take advantage of most of these advances. But if you can provide the business justification, their benefits should be worth the technology investment. For additional information about using Windows 2003 to reduce the complexities and costs of managing your storage infrastructure, see the Microsoft Storage Web site (http://www.microsoft.com/windows/storage).
Many organizations use business drivers to determine which disaster-recovery service level agreements (SLAs) to adopt. The SLAs then determine how extensive the backup strategy needs to be. But even if your SLAs and backup strategy are firmly in place, you might discover that improvements in technology provide good reasons to overhaul or at least fine-tune your strategy. Administrators must reduce backup-and-restore windows while keeping up with ever-growing data sets. To accomplish these seemingly contradictory tasks, you either have to improve your storage capabilities or minimize the amount of data you back up. Therefore, as you reexamine your backup strategy, concentrate on two primary goals: improving the facility and reducing the data set.
Improving the Facility
To improve your backup-and-restore capabilities, concentrate on hardware and software technologies that increase backup-and-restore rates and the facility's capacity. For example, if your servers have one DLT drive that stores 50GB of data on tape at a rate of 10GB per hour during backup, consider technologies that let you back up 100GB at 20GB per hour. You might need to upgrade to a new DLT drive, select a new tape technology, or make the leap to SAN- or VSS-based data cloning or replication. You must also consider how your backup window affects recovery time. If you can back up all your data in 4 hours, you probably have discovered that restore times can be as long as 6 to 8 hours.
Many administrators concentrate on recovery time rather than backup time, which is a smart move. With NAS and SAN becoming more widely deployed and disk-storage costs decreasing (while capacities increase), some administrators are turning to backup-to-disk technology to shrink backup-and-restore windows and are using the technology as a complement or even a replacement for tape-based backups. This popular method has huge advantages over traditional techniques. My favorite procedure is a hybrid approach that leverages backup-to-disk technology for recent backup sets (e.g., T minus 2 days) and stores older (e.g., T minus 28 days) backup sets on tape so that I can rotate them off site. Whichever approach you use to improve your facility, identifying the measures (according to your business constraints and requirements) you can use to more quickly back up and restore more data is of paramount importance.
Reducing the Data Set
Improving the facility and reducing the data set are closely related tasks, but the latter is a more challenging problem. If you increase your facility capability, the need to reduce the data set decreases (and vice versa). By reducing the size of your recovery unit, you can manage that unit within your business constraints and requirements. This scenario doesn't address whole-server recovery but concentrates on the most important part of the equation—the data. One way to manage or reduce the size of your recovery unit is to impose quotas. You could also increase the number of manageable recovery units and thereby reduce the size of each unit. For example, for Exchange Server 2003, you could divide your 50GB Information Store (IS) into five 10GB stores and reduce the time it takes to back up and restore each unit (doing so doesn't actually reduce the amount of data; it simply divides the data into more manageable units). Although this approach increases complexity, it concentrates on manageable units of recovery that affect smaller populations of users.
Data grooming or archiving is another technique that's used to reduce recovery-unit size. This approach lets you concentrate on the most important or mission-critical portions of your data and migrate the less important data to alternative storage locations, such as tape or optical media. When it comes to implementation, data grooming is similar to Remote Storage Service (RSS) for Windows, in which files are migrated to near-line storage according to policies. Exchange and SQL Server also use similar database archival and grooming techniques. But don't be fooled: If you relocate less-crucial data to an alternative storage location, you still need to figure out which level of protection to provide for that data or you could compound your problems. Whichever approach you use, reduce the amount of data in the primary storage location.
Make Your Life Easier
During the past few years, backup software vendors and OS and application vendors have brought us many improvements in hardware and software technology, which should be reason enough to take a fresh look at how you protect your organization's data. If you've already employed some of these new technologies, you might want to think about how you can reduce or optimize your recovery-unit size. If you can maximize your facility and minimize the data you must back up and restore, your job as an administrator will be much easier.