In the next few months, we’ll begin to see sizable flash-based hard drives show up.
They’ll be expensive and relatively slow compared to what we’ll see two-three years from now, but that’s not what’s interesting about them.
What is interesting is what’s not happening in the area these drives are challenging: the rust stuff.
No one seems to give a damn about improving hard drive performance anymore.
Think about it. What was the standard five years ago? 7200rpm drives. What is it now? 7200rpm drives.
Yes, there’s that little 10K drive called the Raptor. Do you realize that the Raptor has been out for four years, and no one else has built a 10K ATA/SATA drive to compete against it?
If the hard drive market were still performance-oriented, what are the odds on that happening?
Yes, today’s 7200 rpm drives are faster than grandpa’s, but the raw performance metrics (i.e. disk transfer rates) show a performance increase of only around 10% a year, and most of that is probably just a by-product of the one thing hard drive are still hellbent on:
Size. The drive makers continue to be plenty interested in that; it’s still expanding at a pretty rapid clip, even though size now vastly exceeds typical traditional computer use.
Of course, today’s big drive aren’t overkill for today and especially tomorrow’s media use. Putting a lot of HD-video on a computer as a media server is going to require multiterabytes of storage.
That’s going to be the foreseeable future of hard drives. As the head of Seagate memorably put it a few months ago, “We’re building a product that helps people buy more crap – and watch porn.”
And since you don’t need or want faster porn (I think), I guess speed is on the backburner.
Which brings us to these solid state devices.
In the long run, I think this is where the action is going to be when it comes to performance.
For the foreseeable future (at least a decade), these drives can’t possibly compete against rust for cost-effectiveness on the media-driven terabyte level. However, these new drives are probably roughly even with rust overall for general computing purposes.
Of course, paying twenty-five times more for essentially the same performance bottom-line is a DOAer unless other factors take precedent, which is why these devices will be found mostly in notebooks for the next couple years.
However, after that, we’ll probably see flash drives that are considerably faster than rust in all aspects, and the cost multiple will be considerably less (and, unlike rust, there will be a close relationship between size and cost).
However, it’s not going to be a matter of one replacing the other. The two will have their separate purposes, and largely go their separate ways, with remarkably little competition between the two.
I think the performance machine of 2010 will have an SSD standard, and rust for the rest. Business machines will follow, if for no other reason than to crimp the style of employees downloading everything but work.
The rest will probably divide into computing and media boxes. The first will go solid-state when the cost of a smallish solid-state drive becomes less than that of a big hard drive; the media (or really cheap) boxes will stay all-rust until/if sticking with rust cripples the machine.