'

Are hard disks on the way out?

Will spinning discs always be with us? That's a question on many people's lips as the price of SSDs continues to tumble, and one I find myself asking as I'm in the process of reviewing a bunch of SSDs for ZDNet UK.

Will spinning discs always be with us? That's a question on many people's lips as the price of SSDs continues to tumble, and one I find myself asking as I'm in the process of reviewing a bunch of SSDs for ZDNet UK.

The general assumption that SSDs will replace hard disks presumes that SSD prices will form an orderly curve, according to the Law of Moore. However, some argue that the limits of SSD technology are already close, as quantum effects start to make themselves felt. While there are replacements for today's NAND technology, they're still immature and not at all cheap. Consequently, 32nm might be as small as today's technology can go -- and it's already at 45nm. So it looks as if there'll be a price-performance floor for SSDs.

But you can't assume that people will switch from disks to chips even if the price of each per gig reaches parity -- if ever it does, which seems unlikely. In fact, you can deploy SSDs in places where the hard disk will never shine, and therefore where the price comparison is close to irrelevant.

For example, enterprises buy SSDs to accelerate servers in datacentres, and install them in places where they'd never install mechanical storage devices for good reasons, such as performance, heat generation and cost of maintenance. In such instances, the price of a hard disk is irrelevant. You also don't need to set up SSDs in a RAID configuration that compensates for the inherent unreliability of mechanical devices, so you would configure them as a RAID level zero for performance reasons. Again, they would not be replacing hard disks - but interestingly, although the cost will remain substantially higher, you won't have to pay the capacity tax that RAID levels 1 and 5 and their derivatives impose, so reducing the cost per gig of the SSD.

Where the price comparison really works is in laptops where, given the choice, most people would opt for a 128GB laptop SSD that uses less battery, is silent, and near-immune to mechanical upset. Users like them for that reason, as do desktop client managers of large companies, who can expect to see lower maintenance costs as a result. Expect rotating media to disappear quickly here.

Yet mechanical hard disks will keep soldiering on, and I'd argue that they'll remain in play for the foreseeable future: they'll remain cheaper than SSDs on a per gigabyte basis, and there'll always be situations where capacity and price are more important than performance. And the hard disk's other Achilles heel, power consumption, becomes a non-issue in some cases, such as in a MAID setup, which effectively switches unused disks off, and the hard disk's power consumption disadvantage against the SSD becomes a non-issue.

I think the home will be the place where the SSD really scores. As broadband speeds increase, the ability to store data in the cloud will mean that local storage needs will plateau. As a result, all you'll need is an SSD disk big enough to boot off - and you can buy one of those now for under 100UKP. Even locally stored data will be stored on chips not magnetic particles. The market for home servers, PVRs and other appliances is immature but growing, and such kit will be widespread over the next five years. In this market, silent running and low power consumption, courtesy of the SSD, will be key differentiators of always-on devices as energy costs rise.

That said, I wouldn't expect that scenario to become mainstream for another five years, during which time rotating media will reign supreme. So in the short, medium and long term, the familiar rattle of the hard disk head is not about to disappear.