Re: Theory vs. reality
Another opinion on big data from a self-interested vendor. Atkinson's "cost millions to data warehouse" perspective is a little dated. And the example he offers, tied to structured transactional data, is also not a very "big data" frame of reference.
The point of aggregating to the hour instead of the second is simple enough -- conventional wisdom, really. But this seems like a very conventional frame of reference focused on developing analytics based on recency, frequency, and monetary value. What about variable data types like clickstreams, log files, or social data? That's when data gets really big. It's not just a matter of collecting more of the same old data.