Data Flow Diagram (DFD)

A data flow diagram (DFD) is a graphical representation of the flow of data through an information system without any indication of time. DFDs are commonly used to provide an initial top-down analysis of a system, identifying the processes to be carried out and the interactions and data exchanges between them. DFDs can be either logical, providing an implementation-independent description of the system, or physical describing the actual entities (devices, department, people, etc.) involved.

Data flow diagrams can be created in a model targeting the Data Flow Diagram process language. To create a new DFD model, select File > New Model, choose Business Process Model as the model type, Data Flow Diagram as the process language, and Business Process Diagram as the first diagram.

Note: To create a DFD model from an analysis or other BPM model, select Tools > Generate Business Process Model and select Data Flow Diagram as the process language. Composite processes with start and end objects are replaced with external entity objects or process shortcuts, depending on the input and output flows of the composite process. Data items are preserved.

PowerDesigner supports all the objects necessary to build data flow diagrams:

Tool	Symbol	Description
	Gane & Sarson: Yourdon:	Process - An activity that transforms or manipulates input data to produce output data (see Processes (BPM)). Flows from processes can go to external entities, data stores, split/merges, or other processes. The Data tab in the process property sheet displays the process CRUD accesses to data. Note: To choose a notation, select Tools > Model Options and select the appropriate Data Flow Diagram Notation.
		Flow - Oriented link that conveys data between processes, external entities, and data stores and represents data in motion. Flows are based on standard flows (see Flows (BPM )) with a `Flow` stereotype. Flows to or from data stores are based on standard resource flows (see Resource Flows (BPM)), and must be created with the Resource Flow tool. A flow cannot directly link two data stores or two external entities. The Data tab in the flow property sheet displays the data transported by the flow.
	Gane & Sarson: Yourdon:	Data store - Location where data resides permanently or temporarily and represents data at rest. Data stores respond to requests for storing and accessing data, but cannot initiate any actions, and are based on standard resources (see Resources (BPM)) with a `Data Store` stereotype. Flows to data stores represent write, update, or delete access, and flows from data stores represent read access.
		External entity - A person, organization, or system outside the system being modeled that sends data to or receives data from the system. Flows from external entities cannot directly access data and must pass through processes. External entities are based on standard organization units (Organization Units ( BPM)) with an `External Entity` stereotype.
		Split/Merge - Splits a flow into two or more flows or merges multiple flows into a single flow. Split/merges are based on standard synchronizations (see Synchronizations (BPM )) with a `Split/Merge` stereotype. A split/merge can, for example, split a complex packet of data into more elementary packets, and send them to different processes or duplicate data to send to different processes, or alternatively merge multiple data packets together for onward transmission.
None	None	Data - Conceptual information exchanged between the other objects (see Data (BPM)) You can associate the data analyzed in a DFD with conceptual, logical, and physical data models and object-oriented models (see Linking Data with Other Model Objects).

Analysts typically begin with a system context DFD to show the interactions between the system as a whole and external entities. In the following example, the Great Care Society process interacts with the Donors, Bank, Members, and Prospective Members external entities:
DFD Top Level Diagram

You can decompose processes in a DFD (see Decomposing Processes), and PowerDesigner initializes the subdiagram with all the objects that link to the process being decomposed (external entities, data stores, and shortcuts to processes as necessary). Such balancing helps ensure that all the flows to and from the process being decomposed are preserved at the next level of decomposition. When the Great Care Society process is decomposed its diagram is initialized with the four external entities:

The diagram is enriched to show the sub-processes that handle donations, deposits, and correspondence, and the data stores that they interact with:

To test balancing, select Tools > Check Model, and select the balancing checks under the Flow and Resource Flow objects.

When using the default Gane & Sarson methodology, processes and data stores are automatically numbered. As you decompose your processes, child processes inherit the number ID of their parent process so that the first child of top level process 1 is numbered 1.1, and the first child of this process is numbered 1.1.1. This numbering scheme enables you to easily identify the lineage of a process at any level of decomposition and provides a convenient way to reference processes and data stores by numbers instead of sometimes long or complex names:

Note: Numbering does not indicate the order in which processes are performed. By default, numbers begin at 1, but you can enter any integer in a process or data store property sheet and all subsequent objects will be incremented from that number. To revert to numbering from one, right-click the diagram background and select Renumber Process IDs or Renumber Data Store IDs.