Setup creates the given tables and fills them with initial data.
The size of the loaded data is returned in bytes, suitable for use with SetBytes of benchmarks. The exact definition of this is deferred to the InitialDataLoader implementation.
Split creates the range splits defined by the given table.
StringTuple returns the given datums as strings suitable for use in directly in SQL.
TODO(dan): Remove this once SCATTER supports placeholders.
InsertsDataLoader is an InitialDataLoader implementation that loads data with batched INSERTs. The zero-value gets some sane defaults for the tunable settings.
InitialDataLoad implements the InitialDataLoader interface.