Reading between the lines, given they're talking > 1 million rows per minute, I'd guess on the order of trillions of rows rather than billions (assuming they retain data for more than a couple of weeks)