Data reacquisition vs. data restoration: a new model for business continuity - Backup/Restore

Computer Technology Review, Nov, 2003 by Diamond Lauffin

For more than 30 years, the backup of fixed disk has been the primary mode of data protection and business continuity. More recently, IT managers have come to recognize that the backup process is important, but that the more pressing issue is data restoration, which is critical when a problem occurs. The need to restore files can be as simple as a single file that was inadvertently deleted by a user from primary disk storage to a catastrophic site disaster on the scale of the destruction of the World Trade Center towers. Whatever the scope of the problem, the traditional approach to file restoration can have a major impact on business continuity.

This recognition of the importance of data restoration has now made the restore window, not the backup window, the primary concern in designing a disaster recovery and business continuity architecture. In effect, the restore window equates to downtime, which can cripple or kill a business in very short order.

The devastating impact of downtime has spurred businesses with unique high availability requirements to deploy active-active server configurations with redundant hardware running in hot or warm failover modes. In this configuration, all data is mirrored between the two servers so if the primary system fails, the standby system will take control with immediate access to data. This type of configuration effectively reduces the restore window to zero by providing an active full file copy of the data set. Unfortunately, these ultimate uptime configurations are beyond the means of most businesses.

With the advent of affordable disk-based backup solutions, the concept of full file format has become significantly more important. The combination of D2D backup arrays using full file format--instead of traditional methods of writing backup data--offers IT managers a dramatic improvement in maintaining business continuity and effectively eliminating the issue of data restoration altogether in favor of a new data protection model: data reacquisition.

The concept of data reacquisition leverages the random access functionality of disk-based backup and writes the backup data in full file format--the same format as the primary disk storage--so that the backup data copy can be immediately accessed without the need for a time-consuming restoration process.

The data reacquisition model is a radical change from traditional backup and restore technologies built around linear storage devices--tape, CD, MO and DVD--that read data sequentially and utilize proprietary file formats, meaning that the backup data files must be restored to disk before they can be accessed by network users.

While the time-to-restore penalty imposed by tape-based backup and other linear secondary storage systems is well known, the full file format functionality is an even more important differentiator in establishing the ability of a backup system to enable users to quickly and easily reacquire data instead of the protracted downtime imposed by traditional data restoration processes.

Amazingly, some vendors of D2D backup systems have opted to discard the advantages of disk-based backup by making the disk systems emulate tape drives, including the use of proprietary data formats. The justification for this approach is that it allows the D2D system to be easily integrated into an organization's existing backup framework: the backup software views the D2D array as just another tape library so the unit can be easily integrated. No changes are required to the current backup software environment, backup performance is improved compared to tape-based backup, and the D2D backup device often provides some level of rapid file restore or recovery compared to tape.

However, achieving these modest improvements creates a huge problem when there is a data loss incident and critical files need to be restored as soon as possible.

Even with "rapid" file recovery features, any backup system that does not utilize full file format must restore the files to disk before they can be accessed by network users. In the case of a hardware failure, the restoration process from the backup device can only start when the faulty hardware has been replaced and the server is up and running again. The time to restore data from tape to primary disk typically takes several hours. That performance can be improved with a disk-based backup device, even one that uses a proprietary format to write data sequentially. However, the time required for the tape-to-disk or disk-to-disk file copy operations is only one component of the overall time to restore. You also must factor in the time it takes to notify an IT administrator of the failure, the time to isolate the failed component(s), the time to acquire and install a replacement, the time to re-boot and re-load system software and set parameters and privileges on the server. Only then can the actual file restoration process begin.

Any solution that can take half a day or more to provide access to critical data is becoming increasingly unacceptable to enterprise IT environments that are focused on data availability and business continuity. The simple reality is that no solution that relies on a 30-year old data recovery model can meet the needs of the typical 24X7X365 environment. To address the current business continuance demands, a new data availability and file recovery model is needed.