The inevitable paradigm shift: disk-to-disk-to-tape. (Tape/Disk/Optical Storage).

The New Reality

Over the past few years, there has been a significant addition to the discussions concerning the backup and recovery of critical enterprise data. Previously, talk in the industry centered on the cost of down time and the hundreds of thousands, and even millions, of dollars per hour that were lost. Today, the conversation still includes the cost of down time; however more priority and attention has been placed on the value of enterprise data and a different set of criteria has been established for assessing risk.

According to industry analysts at the Enterprise Storage Group (ESG), data value depends on its type and who is using it. For example, email loses 95 percent of its value after the first year while database data retains 75 percent of its value. The new reality, which takes into account regulatory mandates and evolving business practices, is forcing organizations to re-evaluate their data protection and recoverability practices.

Market Challenges

Re-evaluation of the value of data sets within the organization is only the first step in understanding the challenges and requirements facing enterprises today. Increasing dependence on data for corporate viability has brought to the forefront the challenges of IT administrators. For example, backup and restore operations have become very cumbersome; the performance advances made by tape technology have not kept up with the needs of business; and tape libraries are prone to mechanical errors resulting in backup failures and, most important, significant delays in restores.

A close look at the current backup environment highlights a number of potential bottle-necks. The first place a bottleneck can occur is the LAN. To minimize bandwidth capacity issues, many organizations have built backup LANs to handle the backup traffic. The second potential bottleneck is the backup application host, which must push the data out; its processing power, BUS capacity, and port capacity will dictate "how fast" and "how much data." On the reception end, a tape library can add tape drives and compress data to increase throughput. Compression (2:1) helps tape drives achieve 30MB/sec throughput per drive; this translates to 108GB/hr, which means that it would take over nine hours to back up a one-Terabyte file with one tape drive. In addition to performance limitations, backups and recovery procedures fail due to failure of tape drives and cartridges, misplacement of cartridges, and accidental destruction of media. This reality would definitely not provide businesses with the confidence that their data is protected and recoverable.

New Approach to Solving Problems

Though many predict that tape will disappear, the reality is that it is not happening just yet. What has occurred over the years is that the role of tape has evolved from being the primary backup medium, to now acting as the archiving medium. This paradigm shift is known as disk-to-disk-to-tape where disk is now being used as a backup solution to augment tape. Below is a quick look at each approach and the pros and cons associated with them.

Each approach has significant implications on day-today operations and effectiveness. As previously mentioned, disk is not going to replace all tape environments any time soon and should be considered to augment tape solutions, providing improved reliability, performance, and availability of your enterprise data. Next, we will examine the different storage backup strategies and how they can solve your particular needs.

Disk to Disk for Primary Backup

Any of the technologies such as snap shot, 3rd mirror copy, tape virtualization, or disk-to-disk backup improves the throughput of data. In the snap shot scenario, data is moved off disk to tape daily in order to protect against device failure and data corruption, as well as to meet archiving requirements. Snap shot is different from 3rd mirror copy in that it saves changes that point to the original data source, so if the original disk fails, snap shot will not help you recover; while 3rd mirror copy is an exact replica of data at that point in time and can be used to recover, in case of failure. Using disk-to-disk from backup application vendors eliminates the need to put data to tape; backup applications use either snap shot or 3rd mirror copy technology. Though both improve the backup process, neither protects against data corruption in the short term and still requires tape backup. With tape virtualization, data is pushed to what the backup application thinks is tape but is actually disk. Once it is on d isk, it can be managed as disk data; native replication applications can he applied for disaster recovery and additional security can be achieved through RAID. In addition to these advantages, tape virtualization is transparent to the backup application and the overall backup and restore process. Using the backup application, operators can still define how long data is retained and when it is time to move it off disk to tape.

Archiving

Archiving can be achieved by sending a cloning command from the backup application to the virtual library. Data is then moved from virtual cartridge to an actual tape library. The operation is controlled by the backup application. Once data is copied from disk to tape, tape can be moved off site. Data remaining on disk can offer improved recoverability for a designated period of time. The diagram shows how a backup application would issue the command, retrieve the data off disk array and copy it to a tape library (see Figure 1).

Disaster Recovery

In addition to archiving, which can be used as a component of a disaster recovery solution since the backup data is on disk, native replication software can be used to move data from one location to another, Once data is at a different site, it can be moved to tape, and recovered for uses such as disaster recovery, staging sites, and development testing (see Figure 2).

With tape virtualization, organizations have the opportunity to take advantage of disk technology while maintaining the manageability of tape.

Backup to disk does address the problem of performance, but it does not address some of the other issues IT administrators face, such as managing data on the backup media, tracking backups from each application or department, flexibility and scalability, and keeping existing processes in place. Of all solutions presented, tape virtualization best addresses these additional needs.

Why Tape Virtualization?

Tape virtualization offers unprecedented improvements in throughput, manageability, scalability, and flexibility of features and functions. Implementation of tape virtualization enables automation of operations, eliminating operator variables such as cartridge handling, loading and storing, offering your budget some operational efficiency. Looking ahead, IT managers can leverage virtualization technology to improve backup processes, continue enhancing performance, decrease recovery point objectives, and eliminate much of the risk associated with the mechanical nature of tape.

Conclusion

The cost of disk has dipped significantly and the prospects of continued improvement in areal density are well known. When all of the potential savings in labor and data availability are taken into consideration, tape is no longer your only choice for backup. It is no longer cost prohibitive to implement disk as your backup repository media, regardless of the size of your enterprise environment, and users should strongly consider this as an option to augment existing tape backup or consider a phased approach where tape will be moved off and used for archiving. As it was pointed out earlier in the article, there are a number of different disk-to-disk solutions that incorporate tape to address the need for archiving and disaster recovery, but only tape virtualization offers the additional benefits of seamless integration, improved management, the ability for the solution to grow organically with business needs, and a total cost of ownership that is appealing even during difficult economic times.

Approach        Pro                          Con

Snap-shot       * Instant point-in-time      * Data resides on
                  copy of data                 production array
                * Instant recovery           * Dependent on primary
                  off disk                     disk for recovery
                * Multiple copies of         * Data is still,
                  data each day                backed up to tape
                * Can be backed up to        * Increases capacity
                  tape without                 utilization on
                  production                   production site
                  data disruption            * Increases capacity
                                               required for solution

Disk-to-disk    * Existing feature part      * Uses snapshot approach
using backup      of the backup application    to backup
applications    * Improves performance       * Requires modifications
                * Higher reliability           to current processes
                * Automation.                  and procedures
                * Improved security          * Limited management
                                               tools exist

Tape            * Improves performance       * Intermediary between
Virtualization  * No changes required to       backup application and
                  existing operations          backup repository
                * Leverages disk drive
                  technology
                * Enables more
                  robust disaster
                  recovery solutions

TLU             * Uses disk to improve       * Data is stored on tape
Virtualization    performance                * Subsceptible to
                * Stores data temporarily      mechanical errors
                  on disk then moves
                  it off to tape

(1)If production disk goes down, snapshot will not help the recovery of
data. This is different from 3rd mirrors, which is an actual exact copy
of product disk at a point in time.

Mike Parise is vice president of sales and marketing and Noemi Greyzdorf is product manager at Diligent Technologies Corporation (Framingham, Mass.)

www.diligent.com