SSDs: Taking the spin out of storage
SSDs remain more expensive than spinning disks. But storage makers are developing clever ways to mix and match them, to cut costs.
Inside the Enterprise: When Apple launched its first iPod way back in 2001 part of its revolutionary nature wasn't its design, or its interface. It was its storage capacity. It could hold the best part of all but the most avid music fan's record collection.
That storage capacity was delivered via a miniature hard disk, originally developed by Hitachi. But, neat though it was, the little spinning disk suffered from all the problems of its bigger contemporaries: fragility, wear and tear, and power consumption.
Apple has long since moved its iPod range to flash memory, solid-state storage. For portable, personal devices that makes complete sense. It's unlikely we would ever have seen an iPhone if Jonathan Ive had had to design it around a hard drive.
The success of flash storage drives has seen them move into ever-larger devices. Tablets use flash storage, of course, but so do Ultrabooks. Power-hungry users from musicians to data analysts are opting for SSDs, not just in laptops but also desktop workstations. And, for very high performance applications, some businesses are using SSD arrays in their server storage.
In servers, SSDs offer the same advantages as they do in personal devices, especially better performance and lower power consumption; they also produce less heat. The reason more businesses don't use SSDs for their storage is simply cost.
It's unlikely we would ever have seen an iPhone if Jonathan Ive had had to design it around a hard drive.
But the standard, consumer flash drive is not that well suited to enterprise storage, forcing businesses that want enterprise performance and reliability to invest in units that cost some four times as much.
According to Bob Fine, storage systems expert at Dell, the difficulty stems from different technologies, designed for slightly different tasks.
Consumer SSD drives are based around multi-layer cell (MLC) technology. Enterprise drives use single-layer cells (SLC) to store data, but SLC drives are more expensive to make. As they are also more specialist, they do not, yet, enjoy the economies of scale of MLCs.
SLC-based storage, though, is designed for the frequent reads and writes of data typical to a server. MLC storage is designed for data that is written less often: perfect for an MP3 player and fine for a laptop, but not for an enterprise storage array.
We could simply wait for the cost of SLC storage to come down. But another solution, Mr Fine says, and one that Dell is pursuing, is to meld MLC and SLC storage and even conventional spinning disks together in one subsystem. Then, use software to fool the server, and its applications, into looking at the array as one single volume.
There are other advantages, too, to this approach: storage software can learn which files, or blocks of data, are accessed most often, and move them from spinning disk to the fastest solid-state drives, and back again.
Tie this in to an information lifecycle management programme where the business actively archives off less used information to lower-cost storage, even including tape and it could be an effective way to control rapidly rising data storage costs, and improve server performance too.
If only it were that easy to find a long-forgotten album and rip it to iTunes.
Stephen Pritchard is a contributing editor at IT Pro.