I just read something that put it all into perspective for me just what that statement meant:
This still doesn't sit well with me - we're not even remotely getting all of this story. A database of numbers that simply say A called B cannot possibly be the "largest database in the world". Virtually every single phone company holds years and years of billing records like these. Even pooled together, they wouldn't create the world's largest database, not by a long shot. If you simply attached an audio file to each of those records, well then now you'd be talking about the kind of dataset that would create the "largest database in the world".
<snip>
1. How "large" would a database be that simply contained every American with a phone line, and a list of all the phone calls he/she received and made over the past, say, five years.
2. What other large databases are there in existence around the world - public, private, you name it - and how would "big" are they?
For example, the IRS clearly has a database of every taxpayer - how big is that, and how would that compare to the database in point 1 above? What other "large databases" are there out there and how do they compare?
http://americablog.blogspot.com/2006/05/we-dont-even-remotely-have-entire.html