The default data caching strategy of Yes
compromises between disk usage and data fetching.
Data for all time series is fetched at the beginning of the flow evaluation and kept on-disk while the flow is evaluating.
Selecting a data caching strategy of No
reduces disk space using during flow evaluation,
at the cost of additional time spent on fetching data from the original source,
as each block in the flow will have to query the source again.
The Smart
data caching strategy minimizes the time spent on fetching data from the original source.
Data is cached between flow evaluations and only new data for the current time range is requested.