The raw LHC data is already filtered down to a tiny percentage (<0.1% IIRC) using several levels of hardware and then software triggers to make sure only the interesting stuff is recorded. So LHC could very easily fill several orders of magnitude more than this.