Choosing the Right Backup Technology

When people think of NetApp, they usually think “primary storage.” While that’s true, NetApp also has a wealth of experience in designing and implementing backup and recovery solutions worldwide, from the largest data centers to small remote offices with limited IT staff. Over 60% of Fortune 100 companies use NetApp for long-term data protection; we have thousands of backup and recovery installations. NetApp has vast experience in helping IT teams choose and implement solutions for difficult data protection problems.
This article examines:

• Today’s backup and recovery challenges
• Different backup technologies to address these problems
• When to choose specific solutions

Key to this discussion is balancing cost constraints against the never-ending need for better performance while leveraging new and emerging technologies. We’ll start by discussing the sources of backup and recovery problems, talk about general classes of solutions, describe selection criteria, and then provide some information on specific NetApp® backup offerings, including NetApp virtual tape library (VTL) solutions and NetApp SnapVault®.

	Pros	Cons
Traditional tape	Familiar operation and technology	Growing complexity, poor reliability
Transitional disk	Improved reliability and recoverability	Difficult to provision, expensive
Virtual tape	Simple to deploy and scale, disk turned for high backup performance	Relies on existing tape processes
Replication-based backup	More frequent backups, long distance replication, storage efficient	Different methodology for backup

Figure 1) Pros and cons of various backup and recovery alternatives.

Traditional Backup and Recovery Challenges

Most organizations, regardless of size, face a set of common challenges:

• Inability to get nightly backups done within available backup windows
• Difficulty recovering data quickly enough
• Reliability issues that cause restores to fail

At branch and remote offices, these issues are compounded by additional challenges:

• Backups often fail to complete
• No reliable off-site disaster recovery
• Staff too busy to handle the complexity of tape storage or backup applications
• Lack of standardization and limited visibility into operations

Tape technology is a significant contributor to long backup windows, long recovery times, and reliability issues that can plague traditional backup environments:

• It can take up to 3 minutes to simply load a tape and position the tape heads to begin reading or writing. And that’s assuming the right tapes are in the library.
• If the backup or restore spans multiple tapes, similar delays are incurred each time a tape is changed.
• If you can’t stream data fast enough to or from the tape drive, it can slow down considerably, leading to “shoe-shining” as the tape drive repeatedly repositions. Most modern tape drives have stepping algorithms to limit this, but there’s still a performance impact that can lengthen your backup window or slow down recovery.

When it comes to recovery, tape reliability often becomes a significant issue that can add delays and uncertainty. Tape is a passive medium. Once you write to a tape, it just sits in a slot in your library until it’s needed again, with no active monitoring to make sure that it’s okay. If there’s an issue with a tape, you don’t detect it until you use it again, and if the next use is a restore, it can significantly extend the time needed to do the restore—if it can be completed at all.

Is Disk an Economical Answer?

It probably won’t come as a surprise to you that adding disk backup to your overall data protection strategy may solve a lot of these problems. Disk solutions increase performance and avoid the load/unload/seek delays that are inherent with tape as well as the throughput penalty for “shoe-shining”.

For recovery, disk solutions largely avoid the reliability problems that can occur with passive tape media. Disks are continuously monitored through the use of technologies such as regular media scans. If a problem does occur, you are much more likely to find it with disk—before you need to restore that critical file or server; and RAID protection allows you to recover from media errors even when they occur during restore.

Naturally, this is where performance versus cost enters the picture. It’s true that a TB of tape is still less expensive than a TB of disk. However, there are ways to mitigate disk costs. Many solutions include compression technologies, and new deduplication technologies can significantly change the cost dynamics of disk versus tape backup.

Compression offers data reduction similar to that found on tape drives, reducing overall capacity requirements. Deduplication uses more advanced algorithms to further increase data reduction. Typical backups have an inherently high level of duplication. Every time you do a full backup, you back up mostly the same files as previous backups, most of which have changed little if at all. Even incremental backups may include many files that have been changed only slightly since the last time they were backed up. Deduplication can yield up to a 20:1 (95%) space savings, significantly reducing the amount of disk you need versus tape, as illustrated in Figure 2. (A companion article in this issue discusses the reliability of deduplication in backup environments .

Figure 2) Deduplicated backups versus traditional tape or disk backups for a typical environment.

Of course, tape may still have a valuable role to play in your data protection strategy; the scope is just narrower than it once was. When you consider that most restores occur within the first week after backup, you can get a lot of leverage from a small amount of disk storage while continuing to use tape for less critical data and offsite storage.

Disk-Based Solution Options

When it comes to disk-based backup hardware, you have two options:

• Virtual Tape Libraries (VTL), which use disk arrays, but emulate the characteristics of tape libraries
• Traditional disk arrays

VTL Advantages
VTLs offer the following advantages; they:

• Fit into your existing backup infrastructure without requiring major changes, making them simple to deploy
• Eliminate the reliability issues associated with physical tapes and libraries
• Are optimized for maximum backup performance
• Offer a simple path to physical tape when it’s necessary to send data offsite

Traditional Disk Advantages
Traditional disk solutions (SAN solutions using LUNs, or NAS) also offer increased reliability and performance, but without the specializations of VTL. Potential advantages of traditional disk include:

• Standard disk interfaces that make it easy to manage data using familiar methods and tools
• Ability to be used for both general disk storage and backup functions (depending on performance demands)
• Enables use of replication-based technologies for disaster recovery and backup/recovery.

Replication-Based Backup and Recovery

The use of replication technologies is also possible with VTL, but it is currently much less frequent due to the need to keep the backup application catalog up to date with the exact position of backup images and all replicated copies. Expect this to be a key area of advancement over the next few years as VTL vendors (and storage vendors in general) increase integration with backup applications to provide additional value-added services (SnapVault for NetBackup™ is one example).

For disk array products, the most common use of replication technology is for disaster recovery. Replication can occur at the array or operating system level but the end result is a synchronous or near-synchronous copy of the data set allowing recovery from system or site-level failures. Although this is not truly backup and recovery (which provides multiple point-in-time images for recovery) this type of replication opens the door for unique opportunities.

The integration of Snapshot™ technology with replication makes it possible to provide high-speed, low-overhead backup and recovery. Implementations vary, but the goal is to replicate changed data from a source system, either at the array or server level, to a target system and then generate a backup image via a Snapshot copy. This provides three key advantages:

• Increased backup and restore performance because less data is moved
• Reduced capacity utilization because only changed blocks, not whole files, are stored as part of each backup
• Efficient WAN backup because only a limited amount of data must be transferred

The end result is a high-performance, low-bandwidth, space-efficient backup and recovery strategy. Depending on the vendor, tape integration may also be available.

NetApp Solutions

NetApp offers a full range of disk-based backup solutions, including two hardware options: VTL and FAS storage systems; and our replication-based backup solution, NetApp SnapVault.

NetApp NearStore VTL
The NearStore VTL product line includes three models. Our flagship model—the NearStore® VTL1400—offers raw capacity up to 672TB (1,100TB compressed) and sustained write throughput up to 2.5GB/sec. All models feature hardware compression, self-tuning for optimal performance, smart sizing for efficient use of physical tapes, and shadow tapes for restores even when physical tapes are stored offsite. Entry-level systems can be upgraded in place to higher models.

NetApp FAS Storage
NetApp FAS storage systems support both high-performance Fibre Channel and economical SATA disks in a single array, allowing you to mix both primary and secondary storage in a single system, or to create SATA-only systems exclusively to support backup. With FAS storage systems, you can immediately leverage all the proven capabilities of the NetApp Data ONTAP® operating environment for protecting and managing your backups, including NetApp RAID-DP®, Snapshot, deduplication, and FlexShare™ for quality of service.

SnapVault
The NetApp SnapVault family is a replication-based backup and recovery solution designed to enhance and simplify data protection, especially at remote offices. SnapVault leverages NetApp Snapshot technology to back up and restore systems at the block level. SnapVault identifies and copies only the changed blocks on a system (not changed files) to secondary storage. This increases performance by limiting the amount of data transferred during backup and restore operations, as well as limiting the capacity needed to store backups, allowing you to perform backups more frequently if desired. On average, customers report a 95% reduction in both backup and restore time. This methodology also gives customers the option of backing up directly from a remote site to a core data center, because only a limited amount of data is transmitted.

For consistent backup and restore of important databases and applications such as Oracle®, Microsoft® Exchange, SQL Server™, and others, SnapVault can be used with the NetApp SnapManager® suite of products.

For non-NetApp storage, Open Systems SnapVault allows you to incorporate standalone servers and diverse storage systems while offering many of the unique advantages of SnapVault. A companion article in this issue discusses the use of Open Systems SnapVault specifically for protecting VMware® environments.

In case you missed previous SnapVault coverage in Tech OnTap, the following recent article highlight case studies of customer deployments:

• A customer-written case study describes Agilent Technologies’ experience in moving to disk-based backup with SnapVault.

Making the Right Choice

Choosing among the available backup technologies can seem daunting, but there are actually a few simple guidelines that will help guide you to the right solution. At the highest level, it’s probably easiest to consider choices in terms of your target environment—data center or remote office.

Data centers typically have a dedicated IT staff with knowledge of backup and recovery. Some level of tape has undoubtedly been in use, and administrators have designed best practices around its usage. In addition, there is probably a mandate for offsite tape storage to meet compliance guidelines. Most data centers are driven by demands from the business units, which require:

• Integration into best practices and an easy way to migrate data to tape
• Scalable performance to meet restore times and backup windows
• Guaranteed reliability for completing backups and restores

To meet these needs, a VTL is the best option. Tuned for backup and providing a traditional tape interface, VTL offers performance, scalability, and integration into existing best practices. Whether through backup application manipulation or value-added technology, VTLs also enable a simple path to tape for long-term offsite retention.

 

You may also want to evaluate traditional disk solutions, depending on what value-added services they provide. Offerings such as deduplication and backup-application-specific data reduction can be extremely compelling for cost-sensitive environments, but be sure to understand their performance capabilities. If the VTL tuning and interface are not as important, this may be the appropriate path. But again, be sure to understand all the performance, cost, and scaling capabilities of any solution.

 

At the remote office, the situation is slightly different. There is often limited or no staffing for data protection, and the key goal is to eliminate as much complexity as possible so that backups simply happen. First and foremost, this means removing or significantly limiting tape to reduce configuration and management overhead, as well as the complexity and cost of media management. You should look at disk solutions not only for simplicity but also for cost-saving features such as deduplication.

Additionally, offices may want to look at augmenting disk backup with replication-based data protection to limit reliance on backup applications. Although backup applications offer deep and rich feature sets, they require a high level of knowledge and deliver more functionality than is really needed—especially if actions like tape offsite and cataloging can be consolidated to the data center. By stripping out the backup application from the remote office and pushing long-term retention tasks to the data center, you can significantly reduce data protection complexity and the chances of data loss as well as violation of regulations. In addition, replication-based data protection can further reduce capacity requirements and speed up backups through the use of block-level incrementals, where only changed blocks are moved and stored. For the most efficient solution, you should consider replication-based data protection that backs up data from remote to core.

If you’ve been hesitant to consider disk-based backup because you thought it was too expensive, it may be time to look again. The latest deduplication technologies can significantly reduce the cost of disk versus tape, and the reduction in complexity can reduce management costs as well.
NetApp offers a full range of hardware and software tailored to help you solve virtually any backup problem. If you never thought of NetApp as a backup vendor before—think again!

David A. Chapa Director, Backup and Recovery Solutions NetApp David has over 20 years of industry experience focusing specifically on data availability, data disaster recovery, and business resumption practices. He is co-author of Implementing Backup and Recovery: The Readiness Guide for the Enterprise and is recognized as an authority on backup and recovery, disaster recovery, and business resumption practices.