I just finished listening to the excellent and highly entertaining web event labeled "Sun's Open Archive Solution" brought to you to Sun's Executive Vice President John Fowler.
One thought that really provoked my mind was the notion of customers requesting to keep data forever. Forever is a long time. And if you like me have a friend with a PhD in Geology, a really, really long time. In our industry, innovation brings rapid change, and more often than not this means that there's absolutely no guarantee that the most critical data you have today can be processed by the technology we have in 10 years, let alone forever. As an example, I have here in my hands a floppy disk written by Wordstar running on CP/M some approximately some 25 years ago. If I where to need this data, I'd have a hard time to get it, and that's only 25 years ago.... hardly "forever".
Therein lies the challenge if you are in the data archiving business, as Sun is with the Sun StorageTek 5800 (aka Honeycomb). How do you implement "forever" ?
If you believe this BBC article, the 5,500 year old pottery piece is the oldest writing of which today's scientists believe to indicate the content of the jar. Think about it. Some archaeologist in the year 7,500 A.D. unearths my floppy disks. What would he believe the content to be ? To give our archaeologist a fighting chance, we need to give him the ability to retrieve the data, to understand the data format, and/or a way to interpret the data. If you as an technology provider keep all this intelligence to yourself, you surely have no right to claim "forever" as your goal post. In that light, for Honeycomb to be open-sourced make perfect sense.
Good luck my future archaeologist.
Posted by Taylor's Take on Sun Storage on February 29, 2008 at 09:52 AM PST #