Brace yourself for 4TB drives next year
geometry again
"At the enterprise level today's arrays with 2TB drives could just double their capacity to seemingly fantastic heights. A downside is that disk I/O isn't getting any faster and the ability to stripe data across spindles to increase I/O rates will become more important."
Striping I/O allows you to balance the I/O better, but it doesn't solve the I/O access density problem which is caused by simple geometry, and the fact that we've got pretty close to the practical limits of disk rotational and seek speeds. Many of us already stripe storage arrays to within an inch of their life and use ever more sophisticated caching software and hardware, but it is rapidly meeting the law of diminishing returns.
Simply put, areal density goes up to the square of linear bit density whilst sequential read and write speeds only go up to linear bit density. Random access is essentially fixed. What this means is that if you quadruple density the achievable sequential read/write access density on a per GB basis is halved whilst the random access density is reduced by a factor of four. To read a 4TB drive from start to end is going to take over 12 hours. If you have to rebuild a RAID set containing several of these then it could take twice that long given that this is normally done "hot".
Given that multiple read/write heads aren't viable on HDDs (unlike tapes), then the only way of fixing this is more, and smaller drives (short of SSDs of course, which is hardly in the same price/GB space as 4TB drives). The random access storage area is in dire need of something fundametnal to break this I/O bottleneck issue (which applies to optical disk storage even more than it does to magnetic HDDs).
