The Data Flow system is composed of a number of subsystems and foundation layers (Fig. 1). The foundation layers are shared by the subsystems, which provide the functionalities necessary to handle the Observation Blocks and Reduction Blocks during their life-cycle. The subsystems include both generic facilities e.g. for proposal handling, pipeline processing, or archiving, and instrument dependent modules and description files (e.g. template GUIs, instrument description files, reduction recipes). More detailed definitions are given in the Data Flow design documents ([1],[2],[3]).
A set of classes is of major importance in the Data Flow system. It includes in particular Templates, Observation Blocks, Frames, Reduction Recipes, and Reduction Blocks. Templates define standard instrument operation modes and configurations. The set of parameters associated to a Template is called the Template Signature. Templates are called in sequence by an Observation Block, which also contains the Phase II information ([5]). The detector acquisition system produces FITS files which are manipulated in the DFS in the form of Frames. Rules are applied on the incoming frames in order to create Reduction Blocks, which contain the set of frames to be processed and a reference to the applicable Reduction Recipe. Reduction Recipes are the data analysis procedures executed to process the frames for the purpose of calibration, reduction, or quality control ([6]). These classes are part of the domain specific layer of the DFS (see Section 3.6).
Figure 1: Software Architecture of the Data Flow system