What solution would you suggest to the problem of long-time secure and reliable data preservation. (And: No, Vint Cerf's «digital vellum» or Alan Kay's "cuneiform tablets of 2015» are not ready or not really applicable solutions IMHO, unfortunately.)
My immediate thoughts:
#1. look to history.
#2. share the data over a WIDE area
how much data are we talking about?
all data. ;-)
No, seriously, in the projects I have in mind that are usually several tens of thousands of images, texts and sound files etc. Mostly "cultural heritage" stuff for which we still have the originals (hundreds to thousands of years old). A typical database size, therefore, could be one or a few Terabyte.
But, I agree with Vint Cerf, that the problem is a general one, effecting all digital data produced. Of course, we do not have to save all of them – archives, libraries, museums… never did –, but for those we regard as valuable, we need a solution. And I don't see one – except, maybe, for a complete new, very well thought-out system of free and open hard- and software [designs], and maintained by an institution that would need to have some "eternal" status like national archives/libraries/museums. This would also be a chance to avoid all the pitfalls of the wildly-grown jungle we're in with all those security problems caused by complexity and solved (we believe) by even more complexity. Or caused by just simply stupid decisions – A20 gate anyone? ;-)