The storage Big Bang began with the first electronic digital computers which used modifiable electronic storage, such as mercury delay tubes, CRT storage, magneto-restrictive delay lines or such. Unlike earlier storage — plugboards and punchcards and papertape — this dynamic storage invited use and re-use. Since capacity was tiny, these early storage devices were accessed constantly. In storage terms this was all hot data, and if we assigned colors to degrees of storage heat, white hot.
Storage was the hardest problem to solve in the development of the first real time interactive computer, Project Whirlwind, the forebear of virtually every computer now in use. The storage problem bedevilled the engineering team until they finally got some storage that was about as reliable as the rest of the system. Again the capacity was tiny, but it was enough to store some programs, so that data wasn’t accessed quite as often and the temperature began to cool.
Now 50 years later data has become much cooler. So much so that even the rapidly growing capacities of disk drives have not outstripped their I/O capacity for most applications (high-performance databases excepted). This is not unlike the expansion of the universe to the point where the few hot spots, the stars, are surrounded by light years of sparsely populated space.
Data temperature will continue to cool, and the techniques for making slow disks seem faster will continue to improve. In the next five years we will see the big iron, “enterprise” storage boxes enter the same market stasis that mainframes now occupy. They don’t go away, and the business will remain profitable for the single largest vendor, but growth will cease.
TagmaStore is not the last of its kind, but there is now a death spiral beginning to see who will be the last vendor standing. I’d have to bet on EMC, but if TagmaStore does what they claim (and we can virtually sure it doesn’t, in all particulars) it or a near descendent may be the last big iron array standing five years from now.