9 Structured design

• Overview of structured design

• Transaction analysis

• Transform analysis

• System integration
 9.3 Structured design

• After SSA has produced a set of DFDs, there are three

steps to turn the design into structure charts (from
which the system can be implemented):

1. Transaction analysis

– divides the system into tractable units

2. Transform analysis

– converts units into structure charts

3. System integration

– recombines charts by linking units together

 Structured design

(a) Complete system DFD from Analysis (c) Transform analysis

creates structure charts

(d) Structure charts

recombined

(b) Transaction analysis divides into smaller DFDs

Transformations of the DFD from analysis into a recombined

structure chart via smaller transaction DFDs & their charts.
Step 1: Transaction analysis

• A transaction has five basic components:

1. event in the system’s environment that causes the

transaction to occur

2. stimulus that is applied to the system to inform it

about the event

3. activity that is performed by the system as a result

of the stimulus

4. response that is generated in terms of output from

the system

5. effect that this has on the environment of the system

Step 1: Transaction analysis

• e.g.,

1. event: student ‘signs up’ for bank account

2. stimulus: information about student, university, date

of opening account, grant source, etc.

3. activity: add account details to bank records

4. response: free gifts to student, chequebook, debit

card, bank statements

5. effect: student can spend money

• Later regrouping of transactions in the system structure

may be as simple as combining with a case statement.
Step 2: Transform analysis

• Goal is to find a module’s core function:

– take DFD model of the problem and transform it

into hierarchical structure

1. First, identify the central transform in the DFD:

• central transform lies at the centre of the input and

output data flows
• sometimes the central transform must be created,
by adding an extra bubble

2. DFD is ‘picked up’ by central transform so other func-

tions hang from it

• central transform forms the main body of the module

3. First-cut Structure Chart is formed

• Bubbles replaced by blocks

• Arrows redrawn to show invocations:

– they describe flow of data in DFD, but they show

flow of control (procedure calls) in Structure Chart
• Data flows added as side arrows (data couples)

2 0
1
1 2
0
3

3 4
4
transform
(a) Unstructured (b) Hierarchical
analysis
DFD problem Structure Chart
adding
description of solution
invocation

Illustration of transformation from a simple DFD to

a hierarchical structure chart describing the solution.
Transform analysis

• Initial transform makes one-to-one mapping of process

bubbles to subprograms

• Central transform is chosen to be the most abstract of

the process bubbles

• Least abstract bubbles tend to be closest to the I/O –

mapped to the lower levels of the Structure Chart

• Need to create a central transform for a procedure that

organizes sequencing of calls to subprocedures (which
is not obvious from a single-level DFD)

• First-cut Structure Chart requires further refinement to

produce a good design
 Step 3: System integration

• Finally, last step combines all separate charts from the

transaction and transform analyses:

update
file
continuation
response
trans− up−
transaction master action dated new
master master master

get valid get update put new

transaction master master master

master for− continuation

matted response
master
format write ask user if
master master to continue

Structure Chart for transaction servicing, including principal data couples

 9 Structured design

• Overview of structured design

• Transaction analysis identifies the types of transaction

in system, used as components of the design

• Transform analysis designs each component separately

by extracting a good structure for it

• System integration combines components of the sys-

tem to give a complete solution
 9.4 Transaction Analysis

• A transaction has five components:

(1) event within the system’s environment

(2) stimulus to the system

(3) activity of the system

(4) response from the system

(5) effect upon the system’s environment

Electrical supplier example

(1) event: George want an electricity service

(2) stimulus: Info about George, his house and installation
(3) activity: Addition of George to customer database
(4) response: A work order to turn service on
(5) effect: George gets electricity

which yields the add new customer transaction type:

1
Turn off date Turn off
customer
service

2
Customer New customer details
Service Add new
Department customer

3
Customer payment Accept
customer
payment
Three transaction types
Transaction structure

The structure chart shows how the components for each of

the transaction types integrate:

Electricity
supply system

Disconnect Add new Accept

customer customer payment
 9.5 Transform Analysis

• procedure for converting components of the main DFD

isolated by transaction analysis into a structure chart:

(1) Draw a Data Flow Diagram of a transaction type

(2) Find the central functions of the DFD

(3) Convert the DFD into a rough structure chart

(4) Refine the structure chart using Structured Design

criteria

(5) Perform verification

DFD for Produce payroll subsystem

Time
sheet PAYE & Employee
records NI tables history
Pay
Time details
report Employee ID
Valid Employee & current pay
1.1 time 1.2 ID & 1.3 slip amounts 1.6
Validate report Calculate gross pay Calculate Create full
time report gross pay tax & NI pay slip
Invalid Employee Pay slip
time net pay details
report
1.4 Employee 1.7
Operating Date details
Create pay Format
system
cheque pay slip
Pay cheque Employee For−
details matted
names & pay slip
1.5 addresses details
Format pay
Pay slip
cheque
printer
Formatted
pay cheque
details
Pay cheque
printer
Structure chart for Produce payroll subsystem
Worked example of transform analysis

1. Draw a DFD

• This should be the result of a structured analysis

2. Identify the Central Transform

• Part of DFD that contains the essential function

• Found by inspection, or by “pruning” afferent (com-
ing in) and efferent (going out) branches in DFD:

(a) Trace each afferent stream from outside to middle,

mark the data flow in its most “rich” form
(b) Trace each efferent stream from outside to middle,
mark the data flow in its most “basic” form
(c) Join all the marks in a closed curve
Update file transaction type:
3. Produce a first-cut Structure Chart

• Structure chart shows flow of control (whereas DFD

shows data flow)

• “Head hunt” for boss module — potential candidates

could come from central transform, one that acts as
a co-ordinator, or create new boss:

if there is a good candidate for boss

(in central transform),
then pick up the boss and let all other
bubbles hang down;
else create new boss and hang central
transform and each afferent and
efferent branch from new boss.
Producing a first-cut structure chart

• Arrow heads removed from data flows — direction of

data flow not necessarily same as direction of call

• Arrow heads for calls need to be added and bubbles

need to be redrawn as square modules

• Names of modules may not correspond to bubble names

— module names depict activities of subordinates too
• Adding read and write library modules gives a first-cut
structure chart:
4. Revise the first-cut Structure Chart
To produce better structure, revise the chart as follows:

• Add read and write modules for accessing sources,

sinks and files of data
• Factor and reorganize the afferent and efferent mod-
ules — keeping the system balanced
• Factor the central transform, if needed.
• Add error handling modules.
• Add initialisation and termination details if required
• Ensure that all modules have names in keeping with
their hierarchical roles
• Show all flags that are necessary on a structure chart
but not on a DFD: e.g., “end of stream” information
• Check all the design criteria and be prepared to im-
prove the design in keeping with those criteria. Look
first at factoring, cohesion, state memory, and cor-
respondence to data structures
• E.g., update file DFD

(1) Use “Match Transaction With Master Record” as

the boss or parent module

(2) Remove data flow arrowheads and names of bubbles,

and lift the whole network by the boss module:
(3) Adding detail gives first-cut structure chart:
(4) Module with worst cohesion is “Update and Put Mas-
ter Record and Report Transaction”

– only has sequential and communicational cohesion

– split it and factor out some functions:
Modifications to design

• All modules are at least sequentially cohesive.

• “Format and Put New Master Record” is sequentially

cohesive and trivial — could be removed, except that
it is used by another module

• “Update and Put Master Record” is sequentially cohe-

sive and trivial — remove it, increasing fan-out of boss

• “Format and Put New Master Record” no longer has

fan-in, but removing it would increase fan-out of boss to
seven and make boss ‘worry’ about details of “Format
new master record” — so leave factored
• Second-cut Structure Chart:
 Summary of Structured design

• Transaction analysis to identify components

1. Event
2. Stimulus
3. Activity
4. Response
5. Effect

• Transform analysis to design components

1. Draw DFD
2. Identify central transform
3. Produce first-cut structure chart
4. Revise and develop design

• System integration to re-combine components