Data storage expert Henry Newman looks back at the history of storage and feels amazed at how often the same mistakes have been repeated. Don’t people ever learn?
It continues to amaze me that we repeat the same errors over and over again. Stupid might not be the best word — or a fair term — but I do love to rant. As George Santayana said, “Those who cannot remember the past are condemned to repeat it.” This time I want to look at some of the evolution of thought, or lack thereof, on storage and storage technology. Why do we want to believe this stuff?
Time Machine: 1991
For those of you who do not remember, or are not old enough to remember, when Seagate introduced its first SCSI drive it had .5 GB of storage capacity. It could read and write about 4 MB/sec and the drive had a seek time of 14 ms and latency time of 6.80 ms. Really nice technology for 1991. Late that year Seagate introduced a 1 GB drive. Wow, now we have two data points for increased density in a year. In 1992, we get a Seagate 2 GB drive with 7 MB/sec of transfer with a seek time of 8 ms and latency time of 4.17 ms. Now we’re up to three data points.
Then it was time for analysts and various large organization that look at technology (and you know who I mean) to draw a straight line to show that the growth rate of storage is now on a path to double say every nine or 10 months. I saw those charts back in 1992 and laughed as I had talked to Seagate, knowing the facts. The first 4 GB drive from Seagate back then — who led the industry — was in 1994 with 9 GB in 1996. So many people in the industry read the analysts chart and believed that disk density would double based on the straight line drawn in 1992. I did not believe it then and I don’t believe it now. But here we go again. Remember the expression, fool me once shame on you, fool me twice shame on me.
Time Machine: 2008
This claim is easily searchable: Flash density will continue to double every 12 to 18 months and flash technology in 2012 to 2014 will be less expensive and higher density than rotating storage (see Why Solid State Drives Wont Replace Spinning Disk ). Both claims I believe were based on charts I have seen come from drawing the same straight lines. Well, we’re more than half way through 2011 and I have seen no signs of flash density or price getting close to disk storage even when comparing low-end MLC to SAS enterprise drives. So why would anyone say such a thing back in 2007 and 2008? Why did we believe it then and not run them out of town? Are we stupid or is it that we want to believe because it fits the model we want? Or is it because it is a simple model?
Read the rest about data storage’s checkered past at Enterprise Storage Forum.