Data Flow Diagrams


1. Use process bubbles to show processing or transformation of data. It is not necessary to show steps where data is not transformed or processed, e.g., reading a record from a file.

2. Data flows must begin and/or end at a process bubble. That is, at least one end of a data flow arrow must be at a process bubble.

3. Show only data flows, not control flows (such as IF-ELSE logic).

The diagram below provides an extreme example of overemphasis on control flows (programming logic). The next diagram shows the correct presentation for the Data Flow Diagram.

4. Avoid black holes (a process bubble that has inputs but no output) and miracles (a process bubble that has outputs but not input).

Naming

Data Flows and Data Stores should receive names that describe the composition of the data involved. (nouns)

Process bubbles should be named using strong, active verbs and objects to stress the basic data transformation or process that is occurring in that bubble.

Fuzzy names for data flows and data stores, and weak or mushy verbs for processes usually indicate a lack of understanding about what is happening within the system. Thus, a periodic evaluation of the names being assigned can serve as a guide to areas where more analysis needs to be done.

Practice Exercise

Locate the errors in the following abstract data flow diagram.