The 411 on raid: when two is always better than one.

If you have been following this column for any length of time, you've no doubt noticed the term "mission-critical application" being used frequently. A mission-critical application is any task which, if not performed, will have a substantial and negative impact on your organization.

There is no application that's automatically mission-critical, though in most organizations, accounting and payroll are pretty sensitive. If your organization performs extensive fundraising, donor and prospect files can be considered mission-critical. In most organizations, though, there is at least one application that's mission-critical, and often a number of them.

Backup is an integral part of protecting mission-critical applications and data. Another approach that you should consider, as a supplement to ordered backup and not a substitute, is to implement RAID disk setups on personal computers that are used in a mission-critical task.

Aside from being the stuff that kills bugs dead, when talking about RAID as a computer peripheral, it's an acronym that stands for redundant array of inexpensive disks. All RAID arrays consist of at least two separate hard disk drives, and can sometimes be as many as five drives.

There are two major reasons for configuring multiple disk drives in a RAID army. The key to RAID is based in the fact that a hard disk is organized in tracks, concentric rings of data on the surface of a disk platter. If a hard drive has more than one disk platter, the drive is also organized vertically, as well as horizontally, with tracks assigned to vertical cylinders.

With multiple hard drives, you can configure the drives into different types of arrays to improve performance, increase reliability, or both.

Faster, Faster

One primary use for a RAID array is to improve disk performance. In a process called striping, tracks are written (or read) consecutively on two drives, first on the primary drive, then on the secondary drive. Because the second drive can be accessed while the first drive is still working, drive access speed is considerably faster. This type of array is called RAID 0 configuration, and the two hard drives appear to Windows as a single large hard drive.

With this type of RAID array, as with most (but not all) RAID setups, both hard drives need to have exactly the same capacity. So if you have two 250GB hard disks in RAID 0 configuration, they will appear as a single 500GB drive to the operating system.

RAID 0 arrays are important when hard drive performance is an issue, such as in transaction processing. There is one major vulnerability with RAID 0--if one drive fails, data on both drives are completely corrupted. While it may be possible to recover some of the data from the undamaged drive, it is difficult, and very expensive.

Another type of RAID configuration, and one more germane to most readers, is RAID 1 configuration, often called "mirroring." RAID 1 arrays can have two differently-sized drives, but the capacity of the array will be limited to the smallest drive in the pair, so it pays to have two identically sized hard drives.

When any data is written on the primary drive, it is also written on the secondary drive. In essence, the secondary drive is an exact duplicate of the primary drive, and if either drive is damaged or suffers from a disk crash, the hard disk controller will automatically switch to the other, undamaged drive and warn you. When you replace the damaged drive, the contents of the mirrored drive will be automatically copied to the new drive, and both drives will continue to be exact copies of each other.

Obviously, RAID 1 represents a significant redundancy in protecting your data and applications, though implementing a RAID 1 array doesn't mean that you should forego other backup arrangements. The downside of a RAID 1 array is cost. The mirrored set of drives appears to the operating system as a single drive, with the capacity of only one drive. To illustrate with the same pair of 250GB drives used in the RAID 0 example, a RAID 1 array of two 250GB drives would appear as a single 250GB drive to the operating system.

RAID 0 and RAID 1 aren't the only way to configure an array of drives. RAID 5 performs striping across multiple drives, but adds additional data in the form of parity information that makes the array a bit more fault-tolerant. The array isn't as redundant as a RAID 1 mirrored array, but with RAID 5, there's a good chance that losing a drive won't crash the entire set of drives.

Finally, there's RAID 10. This requires at least four hard drives, and provides striping across a pair of mirrored drives. If one drive goes down, you'll have to replace both drives in the mirrored array. This type of array provides both data redundancy and great performance, but it requires a lot of drives, room for four drives in the personal computer's case, and a power supply in the PC that can power four hard disks.

Making the grade with RAID

Adding RAID capability is best done when you purchase a new personal computer. Most PCs sold today have the capability to support RAID arrays. You will have to turn on this capability in the computer's BIOS, and then use a combination of keys to access the RAID controller's settings.

Older computers may or may not have RAID capability. The original vendor can tell you if you give them the model number of your system.

When you set the array type and level in the RAID utility, it will format and let you partition the array. This process will destroy any data currently on any of the drives in the array, and will require that you reinstall Windows on the new logical "drive." If you have data and applications that you want to preserve, use a utility like Acronis' True Image to make an image copy as backup, and restore the image to the array when Windows has been reinstalled.

With the low cost of even large capacity hard drives, RAID arrays are definitely worth investigating. Your data and applications will be glad that you did.

Ted Needleman is the former associate publisher and editor-in-chief of Accounting Technology magazine. He is now a technology consultant and writer based in Stony Point, N.Y. His email address is [e-mail]