By Mike Talon
I've heard from many TechRepublic members who have asked me to address disaster recovery options for small businesses. In an effort to meet this request, I'll address RAID systems, which have been around for many years now—offering protection for disk systems and the data they contain.
Depending on whom you ask, RAID stands for either Redundant Array of Inexpensive Disks or Redundant Array of Independent Disks. Both definitions are generally valid, especially when applied to smaller businesses that use self-contained RAID arrays.
While RAID arrays can't prevent software or site failure, they can protect your organization from physical hardware failures at the disk level, which is a very common occurrence. RAID systems are a series of physical disks that act in concert to increase performance and/or protect the system and its data against the failure of any one disk (and in some cases, against the failure of multiple disks).
We can define RAID systems by the type of array in use. Let's look at some of the most common RAID systems.
This involves "striping" the disks, which distributes data across multiple disks but appears to the operating system as a single drive. This level offers no protection, but it does speed up disk operations in most cases. Most hardware- and software-based RAID systems support this type of array.
Also known as disk mirroring, this RAID system duplicates any disk operations to multiple disks. When you write data to what appears to be a single drive, you actually write it to two or more disks.
This level offers data recoverability, but it comes at the cost of performance, since every disk action occurs multiple times. Once again, both hardware and software systems usually support this type of array.
Offering distributed data with parity, this RAID system provides data redundancy and increases speed. RAID 5 stripes data across multiple disks (like RAID 0) while maintaining a system of parity blocks that allow the array to know what data is on each physical disk, even if a disk is lost due to malfunction.
This level requires at least three physical disks. In addition, you lose a portion of your space in order for the system to save the parity data and recover lost data in case of failure.
For RAID 5, you generally need hardware-based RAID controllers. However, most servers come with a RAID controller option, so even smaller shops can take advantage of this RAID level.
and RAID 0+1
These levels are two of the most popular. RAID 10 involves mirroring a striped disk set to another striped disk set. RAID 0+1 stripes and mirrors to an identical set of disks, as opposed to only one disk.
Once again, you'll most likely want to use hardware-based controllers rather than software RAID systems. While these systems offer the best speed and protection, they are also among the most expensive due to the number of disks you'll need to set it up.
Of course, this is just a sampling of the types of RAID systems available to smaller organizations. There are a large number of vendor-specific and intermediate types of RAID arrays, and you should talk with your storage vendor to determine what it can offer.
In addition, keep in mind that Windows OSs can often handle different types of RAID without hardware controllers, so you may be able to create a RAID array without swapping out hardware.
Except for RAID 0, any form of RAID offers same-server data protection against hardware failure. You should supplement all RAID systems with some other form of protection against larger and/or software-based failures, such as tape backup or replication.
However, RAID is a great first step toward disaster recovery planning. RAID arrays offer a lot of flexibility and a great way to protect against hardware failure.
Mike Talon is an IT consultant and freelance journalist who has worked for both traditional businesses and dot-com startups.