Data flow is a new concept introduced in SQL Server 2005. Data flow is a workflow dedicated to data operations. Data streams are also called pipelines. The data flow can be considered as an assembly line, which contains multiple operations in sequence. Each node in the data stream is called a conversion. Data Flow usually starts with source conversion and ends with target conversion. Between the two transformations, the predefined data stream transformations are applied to the data in sequence. Some conversions are synchronous, such as search, conditional splitting, and data conversion. These synchronous conversions can be executed in parallel.
Once the conversion has been applied to a data row, the next conversion can start to process the data row without waiting for the upper-level conversion to process the complete data set. Some conversions are asynchronous, such as aggregation and sorting. These conversions must obtain all rows from the previous output to process and generate output for subsequent conversions.
SSIS Learning (2): Data Flow task (I)
Integration Services Learning (3): Data Flow task (2)
SSIS engineers reveal data streams to you
Writing a custom data flow component (dataflow component) for SSIS: Custom Editor