Walter Scott | AWS re:Invent 2016
summary
Company explains moving a 100 petabytes time‑lapse library into AWS using Snowmobile/Snowball, Direct Connect and Glacier so historic imagery becomes usable instead of stuck on tape. They run machine learning and crowdsourcing on GVX to auto-map 13–15 million structures in Australia, process a continuous inflow of new captures (covering some places daily), and report results that are “orders of magnitude” cheaper and faster than manual surveys while ingesting about 10 PB per year.
AI-written summary · about
