Will the UPS help with this?
 

RAID stands for Redundant Array of Independent (or Inexpensive) Disks. It’s intended to provide performance enhancement (speed) and fault tolerance or redundancy (backup) to a series of hard disks. In the past, it’s been used with SCSI hard drives, on dedicated controllers. It is now cropping up using IDE, and is even integrated into several higher end motherboards.

The most common implementations are RAID-0, RAID-1, RAID-0+1, and RAID-5. Each of these options provides a different set of benefits.

RAID-0 is called striping. This implementation enhances performance by reading and writing data across multiple drives, using 2 to 32 different drives at the same time. This allows the information to be accessed faster by avoiding the bottleneck of sending data to and from a single drive. The downfall of RAID-0 is its lack of fault tolerance. If one of the disks in the array fails, all of the data across the entire array is lost.

RAID-1 is called mirroring, and provides a high level of fault tolerance. A mirror is established by creating an exact copy of one drive onto another. In other words, the same data is copied and stored on different disks so that the data can be recovered from the second disk if the original drive fails. The disadvantage of mirroring is a performance hit when writing to disk, since each piece of data must be written twice.

RAID-0+1 is a hybrid of the previous implementations. Basically, you create a stripe set and then mirror that set to an identical striped set. This provides the performance boost of striping, with the redundancy of a mirror. Users should expect a slight performance hit during the rewriting process of the mirror.

RAID-5, called striping with parity, is the most common and frequently used RAID level. It provides the performance boost of striping, using 3 to 32 different drives, adding a layer of redundancy using parity. Parity spreads the information needed to restore the data on one of the hard drives across the remaining drives in the set. The amount of disk space used to store parity information is always equal to the size of one of the partitions in the set. For example, if a stripe set with parity is created on five disks, each with a 500 MB partition used for the stripe, 500 MB is used for parity information and 2000 MB is available for data storage. Regardless of how many disks are used in a stripe set with parity, data is recoverable only if no more than one disk is lost. If two or more disks are lost, the data is unrecoverable.

Now that we have the levels of RAID established, let’s talk about the power management problems. First of all, the purpose of an UPS is to provide temporary power to the server if the main source of power is unavailable. An UPS should NEVER be used to continuously power a server. An UPS just gives the administrator time to perform an orderly shutdown of the server—to avoid corruption of open data files and to allow users to be warned that the server is being shut down. 

But what if the administrator is not available to shut the system down after a power outage? Fortunately, most high end UPS come with software programs which will perform the shutdown on behalf of the administrator. These packages, based upon the server operating system, will shut down the appropriate services — performing an orderly shutdown. These programs may also be set to run scripts, the most common of which can be used to alert the administrator by pager or e-mail, or to notify users of the pending shutdown. More importantly, this software can be configured using custom scripts, based upon which RAID device is being used, to save any cached information to the array and to dismount (and shutdown) the RAID array. Once these scripts are successfully completed, the server can be shutdown without any issues arising from the RAID array. To learn what specific commands are required for dismounting your particular RAID array, contact the hardware vendor that produced your particular device.