May 08

It has been interesting to read various readers’ input on the subject of data protection over the last number of issues. Data backup must be the most under-rated, least practiced and in my opinion, the most important aspect of IT, bar nothing. I work as a consulting engineer for a data security company called A-Logic, specialising in offsite data backup and redundancy, server systems and biometric access control, and over the years, the general IT community’s reactive approach to data backup has resulted in much data loss and gnashing of teeth, not to mention post-disaster business for us.

One thing that perturbed me reading the letters and articles was the misconception that the various RAID systems constitute a form of backup. This is just not the case. RAID or Redundant Array of Independent Disks, is only a means of improving volume availability in the event of disk failure. The most popular RAID arrays are 0,1 and 5. RAID0 or Striping is when two or more disks are concatenated to form one large volume the size of the sum of the disks, with no redundancy but major performance gains. RAID1 or Mirroring is when two disks are mirrored, providing a volume the size of the smallest disk, fully redundant with little or no performance gains. RAID5 or Distributed Parity is when three or more disks are concatenated to form a volume, with a ‘parity’ the size of one of the element disks spread over the volume. RAID5 provides the best redundancy and performance benefits.

None of these arrays however amount to a form of backup, because should the volume fail by any means other than drive failure, such as data corruption, RAID controller failure, not to mention theft, fire, flood, etc., the data is gone as it is not stored in a separate location.

Data Backup on the other hand is the practice of regularly replicating production data elsewhere so that in the event of data loss, the backup can be retrieved and reincorporated into the production system. The most prevalent methods of backup currently are media backups to a CD/DVD, hard drive, network drive or tape. These methods are primarily manual, written to media which has a limited lifetime, can be damaged in storage or transit, and most importantly, is generally unencrypted meaning that should your backup be stolen or lost, your valuable data is accessible by anyone.

Online backups are a comparatively new but mature technology whereby data is transmitted to offsite storage servers via an IP infrastructure such as the internet. A number of websites come up in a Google search for online backup, but their shortcomings range from non-automation, servers being overseas leading to slow data transfers, a lack of compression, a lack of encryption, or a non-incremental data backup. Incremental online data backup is a technology whereby only the changes in a file or database are compressed, encrypted and sent to the server. This type of backup is best provided by a local vendor who can do onsite restores to minimise downtime.

Whatever backup method you choose, focus on issues such as offsite storage, automation, encryption, and most of all regularity, because your restore is only as good as your last backup.

Tony Bartok