You're right. 1kbit/s over a month would result in 2.47gbit of data transferred, which is a bit less damning. To get to 2.47tbit, you would of course need 1mbit

You're right. 1kbit/s over a month would result in 2.47gbit of data transferred, which is a bit less damning. To get to 2.47tbit, you would of course need 1mbit/s transferred (which is still not that much!). However flawed my example, I think my point remains valid: data usage measured over arbitrary time periods makes no sense when the underlying infrastructure is built with blocks that only care about instantaneous transfer rates.