Data protection: recovery with tape

Rich Harada

In all data protection and business continuance infrastructures (whether for backup, disaster recovery, archiving, or any other data storage protection), tape storage can be used to handle the restoration of the original data when something goes wrong. All data protection and business continuance (DP/BC) systems should be designed with data recovery as the primary goal. The method used to protect the data, such as critical tape backup, is a means to meeting the requirements for data recovery.

In a small office or home office (SOHO) environment, it may be feasible to have a single data protection approach, or a standalone tape drive, to store all data, applications and operating systems that are used to operate the business. But in larger organizations, where there are different departments, locations, and processes, a “one-size-fits-all” approach may be more difficult to achieve. Different methods may be needed to cover the different types of digital information, such as database records, applications, messages, and other office files.

For example: A disk image taken from a “new” desktop can be used to recover the operating system, configuration, and basic applications to the same desktop’s replacement hard disk or to another computer; but a computer that had user-created data or added programs may be best stored on a tape back-up so recovery of the user-specific data along with the system files can be restored.

Additionally, protecting the data on different types of storage repositories (such as mainframes, servers, networked storage, desktops and laptops) may need to be separately addressed but centrally coordinated. Variables to consider when developing a DP/BC system would include the amount of data to be protected, the window of system availability, and the criticality of the data.

Value Proposition

Business objectives, such as the amount of downtime or the amount of lost data that the business can successfully withstand in the face of a disaster, should be the driving forces behind a DP/BC program. Since these protection programs act as your computer’s insurance policies and only provide value when bad things happen, it is also a key business objective to balance the costs of these systems with the value of the data that needs protecting.

Start With the Basics

The foundation for almost every DP/BC program is the use of tape to store copies of important and critical information, both locally and in remote locations, to provide recovery options whenever data is lost due to any reason. Tape protects data from almost any threat or event, and provides reliable storage at high capacities, with fast data transfer speeds at the lowest available cost. Of the many storage components in your data center, tape storage should be part of the total solution, no matter what the specific requirements are for data availability.

“Any backup or disaster recovery system that does not incorporate tape is either not providing total data protection or is costing the organization more than it needs to spend to protect many of its assets,” says Dawn Wortman, Fujifilm’s Data Tape Group marketing manager. “Tape provides great value while meeting the protection requirements for the majority of most companies’ data assets.”

Not All Data is Created Equal

Each type of data is not of equal value to the organization. Not only is the value of data attributed to its application, but also to the number of people who need to have access to it. While you do not want to spend more on a total DP/BC solution than the financial losses you might reasonably expect without one, the only way to achieve this is to tailor specific layers of protection to each data asset class. Spend as little as possible at each layer, assigning only those applications and data classes to the more expensive protection systems as they are truly needed.

Some data sets, such as those found in transactional databases, need to be available all of the time. An inability to access an inventory list or customer account information can have an immediate and detrimental impact on a business’ bottom line by shutting down the order processing flow. In an e-business, the customer will likely go to a competitor if an order cannot be placed at that moment. In contrast, losing a document, an e-mail, a spreadsheet, or even a product design diagram may cause work to have to be redone, but the impact may be less dramatic and less time sensitive, nonetheless important.

For each application, and for each data set that it generates, the storage administrator should determine two main business-related requirements:

Recovery Point Objective (RPO): How much data, measured in time from the point of failure going backwards, can afford to be lost and/or recreated? In other words, how often does the data need to be copied to the DP/BC system? In a standard backup regimen, the RPO might be 24 hours–the maximum time since the last backup was performed.

Recovery Time Objective (RTO): This measure relates to how long the organization can afford to be without the information. Does the business start losing money the second this data becomes unavailable, or after a few hours, or after a few days? Again, it’s critical to balance the cost of the solution against the value of the data.

The Least Common Denominator

In designing a system of any type, where there is a range of requirements that need to be met to satisfy a number of diverse customers, it always helps to start at the least common denominator and then add where necessary to satisfy the more demanding requirements.

“In the case of a data protection/business continuance plan, the common denominator should be an automated tape solution,” says Bob Raymond, StorageTek’s manager of Tape Product Engineering. “Tape can protect any and all data classes, and provides an easy and inexpensive means for transporting data to an off-site location for disaster recovery purposes.”

A tape-based backup program, where tape sets are routinely cycled to a remote location and then returned for reuse, offers the best combination of fast local data recovery (when specific files or systems are lost), remote data recovery (when there is a data center outage), and overall cost. And despite the creative claims made by some vendors in the disk-based backup business, the total cost of acquisition and ownership (TCO) of tape is still a small fraction of the TCO of any disk-based solution. (See “Is it Tape and Disk or Tape versus Disk?”, June 2004, CTR.)

Tape Protection and Data Access

Tape-based solutions may be supplemented by disk-based technologies to improve system performance for those applications that require it. Although tape may be fast enough to move data from point to point, the combination of technologies may provide a better total protection solution.

For example, an application that cannot tolerate any downtime or loss of data would be a candidate for clustered servers and mirrored storage to protect against device failures. A once-per-day backup on tape would not satisfy this requirement by itself. But a disk mirror does not protect against data deletions (when a file is deleted, it is simultaneously deleted from both sides of the mirror), nor does it protect against power outages or other disasters that shut down the data center. Instead, a combination of disk mirroring and tape backup can provide a more complete protection scheme. The mirror can take over if the primary disk drive fails, and the backup tapes can be used to recover specific data files or provide the full backup data set needed if, say, a virus corrupts the entire storage array.

In another example, such as an e-mail server for a global organization, where the availability of time to perform backup operations is extremely limited, a disk-based backup solution can provide for smaller backup windows, and also provide somewhat faster recovery of distinct data files. But again, a disk-based solution does not provide off-site protection and, therefore, normally feed their data to a tape system for offsite transport and longer-term retention to meet compliance mandates.

Lastly, user systems including desktops and laptops need to be protected, especially in businesses that employ a large number of computer users. “Copying data files from network-based user systems to a backup server, which then streams them to a tape system, is the most cost-effective solution to meet data protection requirements for this asset class,” says Jim Milligan, Imation’s executive manager of Tape Strategy. “Using any additional protection layers for this asset class might be a waste of money.”

It may be inconvenient and even a little more costly for a user to wait for restoration of lost information from tape, but for less time-sensitive data, a tape solution is the best option because of its minimal impact on the bottom line–especially compared to the costs of a disk-based recovery option.


Building a data protection and business continuance plan is, for most organizations, a very complicated endeavor. These organizations would be well-served by making tape storage the foundation for their DP/BC plan, the default repository for backing up and archiving data.

Tape continues to be the most broadly based storage solution available to meet data protection, archival and business continuance requirements. Other technologies may be layered on top of tape architecture to provide higher levels of performance to meet the Recovery Point Objectives and Recovery Time Objectives required by some mission-critical applications and digital asset classes, but these technologies complement tape rather than replace it.

Rich Harada is president of the Tape Technology Council.

Opening shots in continuing stories …

COPYRIGHT 2004 West World Productions, Inc.

COPYRIGHT 2005 Gale Group

You May Also Like

Protect Network Security Proactively

Protect Network Security Proactively – Technology Information Eric Ogren Today’s modern, high-performance network demands a securit…

Can SAN Give HSM A Face lift?

Can SAN Give HSM A Face lift? – Technology Information Mitch Cipriano According to Framingham, MA-based industry analysts at IDC Re…

ADIC Scales Unix With Its AMASS 5.0

ADIC Scales Unix With Its AMASS 5.0 – Product Announcement A DIC’s version 5.0 of its AMASS for Unix storage management software enables …

Sassy to scuzzy to sass – I/O with Mark & Hal

Sassy to scuzzy to sass – I/O with Mark & Hal – Column Mark Ferelli Mark: There wasn’t much meat on the bone at the Server I/O Conf…