HDD Power Cravings: 'Green' Hype Or Real Savings?

The impact on HDD (hard disk drive) design due to PMR (perpendicular magnetic recording) and other bit-packing technologies is admittedly of great fascination to me. As far back as 1992 I was predicting that burgeoning hard drive capacities, coupled with application capacity demands that didn’t keep pace, would cause the average hard drive size to shrink over time (common HDD form factors sold today, defined by their platter diameters, include 3.5", 2.5" and 1.8"…flash memory has made smaller-sized HDDs obsolete). And more recently, you can see tangible example of this concept in the form of Western Digital’s 300 GByte Velociraptor, which greatly improves on the capacity of its ‘true’ 3.5" 36, 72 and 150 GByte Raptor predecessors in spite of the fact that it’s a 2.5" HDD surrounded by a passive heat sink that constructs a 3.5" HDD-equivalent form factor:

The other notable impact of increased bit-packing capability, most recently exemplified (for example) by Western Digital’s 2 TByte HDD with a 500 GByte/platter (alternatively stated as 250 GByte/platter side, or 400 GBit/in² areal density) capacity, is its potential to support decreased rotational speeds. Since PMR and comparable techniques squeeze increasing amounts of data into each rotation of a HDD platter, fewer rotations need to occur in a given amount of time in order to deliver a target read/write transfer rate. In fact, it’s fairly common nowadays for a given areal density metric to appear first in the form of a 4,200 or 5,400 RPM drive, with higher-speed 7,200, 10,000 and 15,000 RPM variants appearing later (if at all). Presumably this is because it’s more difficult for the HDD’s read/write head(s) to discern between sequentially stored data bits at higher rotational speeds.

Elementary physics analysis will quickly convince you that all other factors being equal, it takes less energy to spin a slower-slower rotating platter than a faster-churning one (as well as to twirl a smaller-diameter platter versus its more Rubenesque alternative). So it is that WD’s 5,400 RPM HDDs are now known by their GreenPower Technology marketing monickers. Earlier today, the company unveiled an audio/video application-tailored version of its 2 TByte 3.5" HDD with firmware and other attributes customized for high-temperature, always-on environments with extremely random access characteristics that unevenly wear-burden the media. And one month back, Seagate launched its Barracuda LP line, which spins the platters at a non-standard 5,900 RPM rate in attempting to hit a performance-vs-power consumption ’sweet spot’.

Here’s a relevant quote from the draft of my upcoming June 28 cover story on NASs:

More drives tend to equate to more cost, unless you’re comparing, say, a leading-edge 2 TByte HDD against two more mature 1 TByte alternatives in a concatenated arrangement. More drives also mean a larger system form factor and more generated heat, therefore a greater likelihood that you’ll need to employ a noisy system fan. As such, seriously consider 5,400 RPM HDDs instead of 7,200 RPM alternatives. Thanks to dense bit-packing PMR (perpendicular magnetic recording) techniques, the slower-spinning drives will still deliver robust transfer rates, and they’ll consume notably less current.

My comments don’t just apply to NASs, of course. And I can’t help but draw a mental analogy to the situation Intel and its CPU competitors ran into a few years ago, wherein the historical approach to improving performance (cranking up the clock rate) ran into a thermal ‘brick wall’, forcing semiconductor suppliers to redirect their energies into squeezing more parallel processing cores onto a given-sized sliver of silicon…just as PMR squeezes more storage onto a given-sized magnetic platter, thereby potentially negating the need for high RPMs and their associated high temperatures.

‘Good enough’ sustained transfer rates coupled with notably lower current draw and heat generation sounds like a winning combination to me, at least until even more performance-vs-power consumption optimized SSDs hit the necessary price metrics at key capacity thresholds. And it’d give profit-strapped HDD suppliers something else to differentiate themselves with, other than having the lowest-priced product at a given capacity and form factor combination (a transient distinction at best). What do you think, folks…will we not only see smaller form factor drives on average sold in the future, but also slower-spinning ones?