MAID (Massive Array of Idle Disks) provides an alternative to tape libraries that is cheaper than conventional disk architectures. In order to do so, MAID ruthlessly eliminates everything that isn't absolutely necessary for near-line storage. That makes MAID an option storage administrators want to consider, if carefully.
The MAID concept is built around near-line access patterns, which are extremely uneven and irregular. Studies show that half the data stored on tape is never accessed at all. Another 25% will be accessed only once. A fraction of the data, perhaps 20%, will be accessed fairly frequently. Of course, response times to access near-line data are slower as well. The worst case retrieval time for data stored in a tape library can be five minutes or more, but that's considered acceptable, if not ideal.
The most striking difference between MAID and other disk architectures is that most of the time, most of the disks aren't spinning. Typically, only about 25% of the disks in a MAID array -- the ones that contain the data most likely to be used -- are actually moving at any one time. Having most of the disks sit idle adds about 10 seconds to the worst case retrieval time, but it extends disk life and reduces power consumption.
Disk life concerns are important because MAID arrays are typically built with lower-cost SATA (Serial ATA) disks. SATA disks usually have lower MTBFs than SCSI drives and keeping them idle most of the time stretches their life. Power consumption and heat generation are important because MAID products generally cram as many disks as possible into a standard rack. Exavio's Exavault, for example packs up to 3 TB of storage into an array 4.5 cm high.
In spite of the focus on cost, MAID implementation still offers sophisticated features. Copan's Revolution 200T, for example, automatically exercises all disks to test them and has the ability to shift data to spare disks automatically if a disk shows signs of failure.
The first generation MAID products work best as an alternative to high-capacity tape libraries or in speciality storage applications like video storage. In fact the two major products, from Copan and Exavio, are aimed at totally different markets. Exavio's Exavault is designed for the digital media market, including video-on-demand and video archiving. Copan's product is intended as a direct competitor to high-volume tape libraries where the company claims its price of $3.50 per uncompressed gigabyte makes it competitive with existing libraries while offering better performance. Additional MAID products are expected in the next several months, mostly as tape library replacements.
For more information:
About the author
Rick Cook has been writing about mass storage since the days when the term meant an 80 K floppy disk. The computers he learned on used ferrite cores and magnetic drums. For the last 20 years he has been a freelance writer specializing in storage and other computer issues.