Hello,
I would probably make a very poor candidate, but I wanted to ask this - the question gives a 1TB hard drive, and fills it literally to the brim with 4 billion sets of 256 bytes. And then you say that 4 or 5 years ago this relied on some sort of distributed system, but now it can be done on one box. What changed? If the drive is one terabyte, and is used to store exactly one terabyte of data, aren't you going to need to put your database overhead somewhere? Am I completely misunderstanding the question? I find it interesting though. It's an easy problem without the storage space limitation. With it, it's fascinating. Are we thinking to compress the values and/or the keys? Or am I off the mark?