Hamburg - Computer users nowadays have a choice of hard drives for storing data - the computer's drive, an external drive, a memory stick or any of a variety of options.

But 20 years ago, that kind of variety was not available. And the hard drive options of that period cost hundreds of dollars for a mere 20 megabytes of storage. Anything larger was unaffordable.

That spurred research institutes to find a way to set up a giant data storage unit, resulting in the 1987 creation of the Redundant Array of Inexpensive Disks (RAID) at the University of California.

Allowing multiple copies of all data is employed to reduce the risk of data loss. A recent study by Google showed how important it is to back up data stored on magnetic devices.

Focusing on the self-monitoring facility of more than 100,000 hard drive, the study showed that drives with surface flaws tend to generate writing or reading errors within 60 days, a probability level 39 times higher than with drives with pristine surfaces.

Over the years, drives with larger caches became cheaper. Suddenly, RAID systems were no longer important because of their backup capability, but because of their ability to store large amounts of data. RAID systems became more common in servers, but the acronym changed to mean Random Array of Independent Disks.

"It's still of interest for servers," says Klaus Dembowski, an engineer with the Institute for Microsystem Technology at the Technological University of Hamburg-Harburg, who has written about the topic. "But I think RAID is unnecessary for regular desktop use."

Many modern devices, from PCs to network-attached storage devices to multimedia notebooks with two hard drive slots sometimes employ RAID technology, most advertising the presence of Level 0 or 1.

The level designation describes exactly how the drives cooperate. Level 1 indicates mirroring, in which the same data is saved on each drive. This is supposed to offer the stronger security against data loss. Level 0 indicates striping, which creates the possibility of storing more data, since the information is spread out across a single drive.

Gamers tend to focus on Level 0, hoping that it will boost their processor speed. "But if the drive isn't good, it won't work," says Dembowski. Additionally, if you only have one drive and experience a memory loss, you lose everything.

But that does not mean Level 1 is necessarily safer. True, the data is repeated on both discs, which is useful if one drive suffers a complete memory loss. But if the user makes a mistake or accidentally deletes data, the information is wiped from both drives.

Thus, RAID is no replacement for a true backup.

"If the point is to get increased security, then you should make a backup," says Dembowski. That means using a separate storage device that is not regularly connected to the main device. USB drives or memory sticks are recommended.

Indeed, RAID technology only really begins to show its stuff at Level 5, which offers more storage combined with redundancy. However, the technology is nearly impossible to find in standard consumer devices.

"RAID 5 combines the advantages of both systems for the high-end of the market, by spreading data across three discs, any of which can be recreated using the other two in case of a loss," explains c't, a German computer magazine.

INFO-BOX: RAID often wrong for private use

Separate controllers usually handle the coordination of RAID hard drives. Those controllers appear as a single logical drive to the operating system. But in consumer devices, the PC processor often has to take on the work of the controllers, which are left out for cost savings reasons. That, in turn, raises the risks of data loss, precisely what RAID is supposed to prevent.

Thus, RAID systems without their own controllers are sometimes termed Fake-RAIDs. If given the option, when setting up their RAID array, users should pick the option Just a Bunch of Discs (JBOD), advises c't. That ups the odds that, in case of a drive failure, the information will be available on an alternate drive. (dpa)