DATA FLOW DIAGRAMS

What is a DFD
In the late 1970s data-flow diagrams (DFDs) were introduced and popularized for structured analysis and design (Gane and Sarson 1979). DFDs show the flow of data from external entities into the system, showed how the data moved from one process to another, as well as its logical storage. Figure 1 presents an example of a DFD using the Gane and Sarson notation. There are only four symbols:

Data Flow Diagrams

Squares representing external entities, which are sources or destinations of data. Rounded rectangles representing processes, which take data as input, do something to it, and output it. Arrows representing the data flows, which can either be electronic data or physical items. Open-ended rectangles representing data stores, including electronic stores such as databases or XML files and physical stores such as or filing cabinets or stacks of paper.

Systems Analysis
Focus is the logical view of the system, not the physical What the system is to accomplish, not how Tools: data flow diagrams data dictionary process specification entity-relationship diagrams

Data Flow Diagram:

"a network representation of a system. The system may be automated, manual, or mixed. The DFD portrays the system in terms of its component pieces, with all interfaces among the components indicated." - Tom DeMarco hence DFDs: focus on the movement of data between external entities and processes, and between processes and data stores

Data Flow Diagrams

n the late 1970s data-flow diagrams (DFDs) were introduced and popularized for structured analysis and design (Gane and Sarson 1979). DFDs show the flow of data from external entities into the system, showed how the data moved from one process to another, as well as its logical storage. Figure 1 presents an example of a DFD using the Gane and Sarson notation. There are only four symbols:

Example Data Flow Diagram

data store

data flow

process

external entity

Sources/Sinks
(external entities)
Any class of people, an organization, or another system which exists outside the system you are studying. Form the boundaries of the system. The system and external entities exchange data in the form of data flows. Must be named, titles preferred to names of individuals - use a noun

source / sink

Data Flows
data in motion marks movement of data through the system - a pipeline to carry data connects the processes, external entities and data stores Unidirectional originate OR end at a process (or both) name as specifically as possible reflect the composition of the data - a noun do not show control flow! Control flow is easy to identify- a signal with only one byte - (on/off). HINT: if you can't name it: either it's control flow, doesn't exist or you need to get more information!

Processes
transform incoming data flows into outgoing data flows represent with a bubble or rounded square name with a strong VERB/OBJECT combination; examples: create_exception_report validate_input_characters calculate_discount

process

Data Stores
data at rest represents holding areas for collection of data, processes add or retrieve data from these stores name using a noun (do not use file) only processes are connected to data stores show net flow of data between data store and process. For instance, when access a DBMS, show only the result flow, not the request

data store

Data Flow Diagram Donts

process stuff

1. Black Hole

process stuff

2. Its a Miracle

Data Flow Diagram Donts

ds-1 A.1 A.2 data

4. Leave Things Unlabeled Corollary: Labels Should Have Meaning

Data Flow Diagram Donts

data store
5. Miracle data source

data store

5. Black hole data source

Data Flow Diagram Donts

6. Data Flows Unassociated With a Process
entity to entity

data store to entity or reverse

data store to data store

Data Flow Diagram Donts

1. 2. 3. 4. BLACK HOLES MIRACLES Let it get too COMPLEX: 7 2 processes Leave things UNLABELED (corollary: labels should have meaning) 5. Data stores that are SOURCES or SINKS 6. Data flows that are UNASSOCIATED with a PROCESS 7. Expect your diagram to be perfect the first time!

Now you try a DFD

Attempt a DFD of your own draw a DFD for the enrollment process at CFBC. Draw it on paper. Review your DFD Dos and Donts to make sure you are not violating the rules. Then use Visio or Microsoft Draw Tools and draw the DFD you jotted down.

Data Flow Diagrams are:

Used to perform structured analysis to determine logical requirements A graphical tool, useful for communicating with users, managers, and other IS personnel Useful for analyzing existing as well as proposed systems A relatively simple technique to learn and use

Why Conduct Process Modeling?

Understand components of current logical or physical system for purpose of rebuilding in a different physical form/technology, possibly with some changed functionality Find inefficiencies in current system Re-engineer current system

Diagramming A System
multiple DFDs are required to represent a system DFDs are created at increasing levels of detail

Different Types of DFDs

Context diagram Level-0 diagram (system diagram) Level-n diagram Primitive diagram

Context Diagram
defines the scope of the system by identifying the system boundary contains: one process (which represents the entire system) all sources/sinks (external entities) data flows linking the process to the sources and sinks (external entities)

Example Context Diagram

student
course selections

schedule

Registration System

Registration details

business office

Constructing a Context Diagram

identify and list sources/sinks (external entities) identify and list inputs to and outputs from sources/sinks (external entities) create context diagram

Level-0 Diagram
describes the overall processing of the system show one process for each major processing step or functional requirement data flows from the context appear on system diagram also (level balancing) can show a single data store to represent all data in aggregate at this level can draw duplicate sources, sinks and data stores to increase legibility

Drawing a Level-0 Diagram

list the major data stores list major business steps draw a segment for each business step assemble into single DFD re-organize until satisfied number processes

Functional Decomposition
similar to a series of more detailed maps iterative process of breaking the description of a system into finer and finer detail to create a set of charts in which one process on a given chart is explained in greater detail on another chart referred to as exploding, partitioning, or leveling must use your judgment to decide what goes on each level show error and exception handling on lower levels (if at all)

Lower Level Diagrams

explode the processes shown on the level-0 diagram each process is represented by its own DFD balance data
data flows on upper level appear on lower level, or data flows on upper level are broken into component pieces with components shown on lower level

each lower level shows greater and greater detail follow numbering convention

Balancing DFDs
conserve data from level to level inputs and outputs on the higher level must appears somewhere on the lower level

Advanced Rules
Composite data flow on one level can be split into its component data flows on the next level - but new data cannot be added and all data in the composite must be included in the sub-flows The inputs to a process must be sufficient to produce the outputs. Lowest level DFDs may add new data flows to represent exception handling, i.e., error messages May repeat data stores or sources/sink to avoid crossing lines

Additional Guidelines
the inputs to a process are different from the outputs of that process objects in a set of DFDs have unique names do not change data flow names on lower levels unless you are decomposing a data flow into component pieces. never explode a single process into another single process. If you cannot partition the process, then the lower level DFD is not needed. expect to iterate, put down the DFD and go back to it a few times to create something satisfactory.

Other Questions about Lower level diagrams

1. How deep? (how many levels?)
if the process has only one input or one output, probably cannot partition further; can you describe the process in English in about 1/2 page?

2. How broad? (how many processes on a level?)

7 two is a reasonable heuristic may temporarily place much of the system on a single diagram then re-draw into separate levels

Quality Guidelines
Completeness
all components included & in project dictionary

Consistency
between levels: balancing, leveling

Timing considerations
assume system never starts and never stops

Iterative nature
revisions are common

Drawing primitives (lowest level)

when to stop?