It's time for 'instant restore appliances'

The goal is rapid recovery of data and applications, not moving and copying data more quickly.

By Kirby Wadsworth

For years, companies have relied on traditional data-protection techniques—tape backup, mirrored disks, periodic mirror break-offs, remote replication, and snapshots. However, each of these techniques falls short of delivering what companies truly need: rapid business recovery.

Kirby Wadsworth
Click here to enlarge image

Newer technologies such as faster tape, disk-to-disk backup, automated break-offs, compression, and transmission caching have improved the ability to copy and move more data more quickly.

However, the business problem is not, and never has been, a need to copy and move more data more quickly. The business problem is rapid recovery of business applications in the event of data loss.

If the business problem is getting data back to a previous, known good state, why do we need to copy and move it?

Until recently, the storage industry's ugly little secret has been that users must rely on "move and copy" data protection because moving and copying data is all the industry knew how to do.

Resolution of this issue is fundamental. Today's business data is massive—terabyte-level databases are prevalent. In the face of large-scale data loss or corruption, application recovery windows have shrunk to minutes. In financial services, for instance, the federal government is demanding full business recovery in two to four hours.

It's time for a new way of thinking about the restoration of data and subsequent recovery of business applications.

What's needed is a new class of storage device that restores data without move and copy processes. This requires the ability to instantly restore to any previous point-in-time, ensuring rapid application recovery and timely resumption of business-critical operations. Such a system would offer user benefits such as

Reduced business risk—Without rapid restoration, business functions cease, customers may be lost, regulations may be breached, and system-wide failures can result. Instant restore technology allows data and business applications to be recovered in minutes, rather than the hours or even days that have been the norm.

Reduce costs—Today, IT attempts to mitigate business risk by over-provisioning storage; sometimes companies have five or six copies of their critical data online at any one time. Instant restore technology allows companies to re-allocate expensive high-

performance disks to online applications and delay new hardware purchases. Plus, by reducing expenses, such as tape cartridges, archiving costs, and networking costs, this type of system could pay for itself within months and deliver long-term ROI.

Simplify the data-protection process—Today, data protection requires complex mirror splits, backups, tape ejection, and archiving. As the underlying data changes, administrators must constantly monitor backup systems to ensure proper configuration. Often, even with the best technical support and procedures, recovery still fails. By simplifying data protection, instant restore frees IT resources to focus on other critical initiatives, in addition to improving recoverability.

Inexpensive silicon and storage capacity are enabling a new type of storage device, which I'll refer to as an "instant restore appliance." This type of device leverages processing power to add time as a new dimension to storing data. Now data can be stored as more than ones and zeros. It can be stored as ones and zeros at an exact moment in time. When asked to respond to a read request, the restore appliance can just as easily respond with data in its present state or as it existed two minutes, two seconds, or two weeks ago.

In the event of a media failure, data corruption, or any other event that would require a restore operation, administrators can request data from a specific point-in-time prior to the event and receive it instantly. Unlike mirror breaks, which only provide the data from scheduled recovery points, an instant restore appliance lets administrators select arbitrary recovery points right up to the exact point of failure.

These appliances could eliminate the backup window without the overhead and cost of multiple mirror sets or snapshots. And, because the presentation of time-stamped data is non-destructive, users can recover an application quickly without compromising the failure history, enabling after-the-crisis analysis.

Today, the laws of physics limit the ability to move and copy data over long distances. The speed of light is fast, but not fast enough. Propagation delays over 30,000m make it impossible to move and copy data synchronously between sites that are far enough outside regional recovery zones to be approved for redundancy. Today's approach is rolling synchronization—breaking off mirrors as quickly as they are fully synchronized on the far-end of an expensive network circuit. This technique requires lots of expensive online disk capacity. It also locks users into a specific vendor and can leave hours of vulnerability to data loss.

Instant restore technology offers a cost-effective solution to the problems associated with longdistance data replication. The restore appliance's "time memory" resolves the write-order consistency problem and allows appliances connected over low-cost network connections hundreds of miles apart to instantly restore to a known good point-in-time just prior to disaster.

If New York goes offline at 11:15 AM, administrators in Chicago can instantly restore data to 11:05 AM, restart the database and applications, and have the business operational in minutes rather than the days required to load tapes in a truck, drive them to Chicago, restore them to scratch space, and implement recovery.

The technology required for instant restore is available today from a number of vendors. (For more information, see "Focus shifts from backup to recovery," p.1.)

No organization is going to rip and replace existing storage infrastructure or trust critical data to an untested technology, so instant restore appliances must complement existing technologies, not replace them.

By adding the dimension of time to the storage of data, instant restore technologies will allow IT organizations to fundamentally change the way they implement data protection in the future.

Kirby Wadsworth is vice president of Revivio Inc. (www.revivio.com) in Lexington, MA.

This article was originally published on July 01, 2003