Intelligent ATA arrives to meet backup and recovery SLAs.

All organizations, regardless of size, are concerned with data protection. Yet few have adequate backup and restore processes in place to ensure they can (1) protect their data, maintain business operations and user productivity, (2) accommodate exploding data growth, and (3) ensure compliance with emerging government regulations (e.g. Sarbanes-Oxley).

IT managers have turned up the heat on storage administrators, requiring their strict adherence to service level agreements (SLAs) that are impossible to meet. Evolving SLAs dictate shorter backup windows to reduce departmental downtime, while government regulations dictate fast and ongoing access to more and more data. Unfortunately, though, many organizations report failure when it comes to successful backup and recovery. In fact, confidence in the backup and recovery process is currently the lowest of any activity in the data center.

How can organizations ensure adequate data protection and regulatory compliance in our highly volatile world? Through a combination of new and evolving storage technologies, as well as careful consideration about where and when to deploy those technologies, IT managers can give their organizations the tools they need to guarantee success in this critical data center activity.

[FIGURE 1 OMITTED]

First, organizations must develop a formal backup and restore policy that:

* Balances the need to backup all of the data as frequently as possible, against the cost of the solution and the outage time required

* Ensures that a history of backups are securely sent offsite in the event of an onsite disaster

* Creates a methodology to periodically test all archived backup media for reliability and ensures data integrity on that media for complete and full data recovery

There is a variety of emerging backup and restore technologies that can help storage administrators successfully execute against their policies. Emerging disk-based backup solutions can significantly reduce backup and restore times, while giving organizations regular and fast access to data. In addition, they can be augmented with a smaller tape infrastructure for offsite data rotation and archiving.

Backup Technology Considerations

The two key elements of backup and restore are the recovery time objective (RTO) (i.e., how quickly the data must be restored) and the recovery point objective (RPO) (i.e., how current the data must be once it is restored). There are some applications (e.g., stock transactions and other online transaction processing functions) where recovery time is as little as one hour and each transaction must be current. This, therefore, requires that each and every I/O be backed up in a synchronous manner. For such applications, monolithic Fibre Channel solutions with disk-based replication features are currently the only way to go. However, for the majority of backup and recovery environments, monolithic Fibre Channel disk arrays are overkill.

On the reverse end of the cost equation, until now low-cost tape had been an attractive alternative to more costly disk-based solutions. While tape still has its place in the data center, companies are under increased pressure to ensure regular and fast access to data in order to comply with new governmental regulations. If a company finds itself under scrutiny by the SEC, it will be given a limited amount of time to produce required data. Sifting through thousands of tapes may not be feasible. Moreover, the unreliability of tape means there are no guarantees that the data required would actually be there.

[FIGURE 2 OMITTED]

In a survey of 500 IT departments completed in January, Meta Group found that as many as 20% of routine, nightly backups fail to capture all data. Compounding the problem, the only way to verify that data has actually been successfully captured on tape is by manually reading the data on the tape. Tape management complexity and reliability issues mean that over-reliance on this storage media could spell disaster for many organizations ("Concerns Raised on Tape Backup Methods", Keith Regan, www.searchsecurity.com, 4-15-04).

Intelligent ATA: An Exciting New Option in Disk-Based Backup

Fortunately, users have another backup and restore technology option in the form of low-cost Serial ATA disks. However, not all Serial ATA solutions are created equal. In fact, many customers who have deployed Serial ATA solutions report that backup and recovery time remains a key issue. Many only deliver around 50-70 MB/sec of performance, equivalent to only 2-3 LTO2 tape drives. Certainly, these solutions reduce management complexity and make it easier to verify successful backup, but nothing has been done to close the growing gap between shrinking backup windows and growing amounts of data.

Traditional SATA architectures aren't delivering the performance needed to handle the 30-plus days of data that 55% of users want to keep on disk-based backups (The Evolution of Enterprise Data Protection: User Adoption of Next-Generation Backup and Recovery Technologies, Enterprise Storage Group, December 2003). Moreover, they are challenged in their ability to read a backup while writing an archive to tape. Even modular array solutions, whether built using Fibre Channel or Serial ATA drives, will suffer performance degradation because they are limited in the number of FC loops that can be supported for storage.

Today, however, there is growing excitement around a new class of Serial ATA product known as Intelligent ATA. The solution combines an intelligent network storage controller with Serial ATA disks in a comprehensive, off-the-shelf modular Serial ATA array. Intelligence enables the solution to deliver the same enterprise-class features as an expensive, proprietary monolithic Fibre Channel device, while Serial ATA drives remove the hefty price tag.

Similar to the design of a monolithic Fibre Channel solution, Intelligent ATA solutions feature independent elements for host connectivity and disk processing, which can be added incrementally to scale capacity and performance (Figure 1). Aggregating network storage controllers in the center of the system are key to its success. Since individual ATA arrays are RAID controllers, the ability to create a single physical pool of storage and spread a single data volume across multiple ATA controllers is critical.

Distributing the load across the distributed cache and RAID processing allows performance to scale with capacity since there are never more than 16 drives behind a single controller in typical configurations. Contrast this to a conventional modular array with a limited number of Fibre Channel loops where there is sure to be performance degradation as more ATA drives are added to the loop (Figure 1). Intelligent ATA array performance can scale as capacity grows into the hundreds of terabytes. Other solutions may leave you exposed, adding capacity without adding RAID processing power.

Performance data shows that Intelligent ATA solutions can deliver up to 2 TB/hour of read/write backup capability (450 MB/sec of performance), which is equivalent to 12 to 15 LTO2 tape drives. This performance makes Intelligent ATA an attractive, low-cost option for satisfying evolving SLAs. Without intelligence, Serial ATA drives have trouble scaling beyond 8TB. With it, performance can scale into the hundreds of terabytes without performance degradation. Intelligent ATA will be a necessary element of an effective backup and recovery process, delivering the performance required to reduce backup and recovery times in disk-to-disk-to-tape backup applications (Figure 2).

The Changing Role of Tape In the Data Center

While Intelligent ATA is a central element of an effective disk-to-disk-to-tape solution, there are benefits to tape-based media that will give it some staying power over the short-term. Tape is very inexpensive; however, as users attempt to get better performance and reliability from tape, they tend to overbuild their tape infrastructure for peak load requirements during backup. For these reasons, it is best to use tape in applications where performance and backup time are not key requirements.

As a result, tape is best used as an archive medium that enables organizations to maintain a history of offsite backups. As removable storage, tape can be easily transported out of the data center for safe keeping off site. New, solid write-once, read many (WORM) tape technologies give users a solution that ensures data can be maintained in a non-rewriteable, non-updateable format--yet another requirement of emerging government regulations.

With an organized history of tape archives, data can be protected from corruption by viruses or operating system problems that can lead to lost data. Moreover, tapes can be easily shipped and are less likely to be damaged in the shipping process. Disk drives must be carefully packed and protected, while tapes can simply be thrown into a box and sent off to their destination. Of course, this also means they can be easily removed by virtually any person in the organization, so companies must have a secure location and checkout process for handling their tape backup.

Backup is Simply a Means to an End

The sole purpose of backing up data is to ensure fast and effective data protection. Organizations must give careful consideration to their data protection needs, understand the impact of data loss, and determine backup and restore policies that accurately correlate to the business-importance of their data.

While mission-critical industries, for the foreseeable future, will continue to require the disk-based replication features only satisfied by expensive monolithic Fibre Channel solutions, this is overkill for the majority of disk-based backup needs. For the majority, Intelligent ATA delivers the performance and reliability currently not satisfied by tape, standalone Serial ATA, or even modular Fibre Channel solutions. As a result, it gives users access to a reliable and fast backup option that is easy to manage, while being significantly less expensive than monolithic Fibre Channel solutions. In addition, it provides a complementary technology to tape, which remains an effective archive solution.

Whereas, in the past, customers were often forced to choose between suboptimal solutions, today there are broad choices in the kinds of storage technologies and solutions that can be deployed. This should lead to greater success in conducting effective backup and restore, and ensuring that business-critical data is always protected.

Stephen Terlizzi is vice president of marketing and business development at Candera, Inc. (Milpitas, CA)

www.candera.com