• large amount of data.
• mostly at the end of some pipeline/historical data // storage + analytics
• read access.
• mostly an analytics thing where you want to draw patterns out of this massive corpus of data

references: