Eliciting Coordination Policies from Requirements

1
Eliciting Coordination
Policies from Requirements
Henry Muccini Fabio Mancinelli SEA group,
Computer Science Dep., University of LAquila,
Italy muccini,mancinel_at_di.univaq.it
2
The Framework Kluwer03 steps

• Elicit coordination requirements
• Map coordination into Software Architecture (SA)
• Analyze consistency between Coordination req. and
  SA
• Mapping this information into the low level

drives
drives
1
Software Architecture
Software Architecture
2
4
Requirement Engineering
Design Coding
Coordination
Coordination
3
Coordination
validates
Rational Unified Process
3
This paper

• We analyze the first step
• How to drive the coordination properties
  elicitation process
• from requirements
• Using the Rational Unified Process (RUP)

RUP-based models
(Informal) Requirements
Coordination elicitation process
4
Something on the RUP

• Model-based process
• Use case based, architecture centric and
  iterative
• Main goal to use UML diagrams to model
• requirements (use case model),
• analysis (analysis model),
• design (design model) and
• implementation (implementation model).

5
Example (a very simple one)

• Students and Professors want to run some tasks by
  using available resources

Use case model
Analysis model
Student
UCC5)
boundary class
6

Based on

• Malone and Crowston
• Actors, Process dependencies and resource sharing
• Arbab
• The need of an exogenous coordination model
• Fiadeiro et al.
• Coordination contracts as connectors
• Linda and Gamma
• What to coordinate, how to coordinate
• Papadopoulos
• Software architecture description implemented
  through coordination languages

7
Our perspective RUP Coordination
Actor
Task

• In terms of Actors, resources and Use Cases,
  coordination may be defined as
• the rules dictating how multiple Use Cases,
  initiated by Actors, can collaborate by utilizing
  available resources (design-based coordination)
• (SE view of coordination)
• From many definitions, to coordinate software
  systems, three entities must be taken into
  consideration
• Functional entities (processes, tasks or
  activities)
• Resource entities (accessed by functions)
• Agent entities, able to run functions

Resource
8
What we could do
Actors
Tasks
Resources
Printer
UC(1)
Printer-fax-copier
UC(2)
Student
Student
Student
Fax
UC(3)
UC(4)
Professor
Professor
How do students and professors can run many print
and fax tasks using the available resources
9
What we propose, instead

• Incremental approach
• Structural and static coordination
• Identify what we need to coordinate
• Identify coordination policies (the how to)
• Based on the concept of Process coordination and
  Resource coordination
• Exogenous model

10
Step0

• Static Coordination (what is need to be
  coordinated)

entity class resources
control class coordinators
11
Step1

• How resources may be allocated to Processes/Actors

Printer
PrintCoord
Printer-Fax-Copier
Printer
Print
Printer-Fax-Copier
PrintCoord
Fax
Printer-Fax-Copier
Fax
12
Step2

• How different Use Cases (of the same Actor)
  interact

Printer
Print
Printer-Fax-Copier
Printer
Print
Printer-Fax-Copier
queue
Fax
Fax
13
Step3

• How different Actors interact

Printer
Print
queue
Print
Printer-Fax-Copier
queue
Fax
Fax
14
Step4

• How many Use Case instances cooperate

Printer
queue
Print
Printer-Fax-Copier
queue
Fax
Print
Fax
queue
15
Step5

• How many Actor instances cooperate

Printer
queue
Student
Student
Student
Print
Printer-Fax-Copier
Fax
Print
queue
Professor
Fax
Professor
queue
16
What we propose, instead

• Incremental approach
• Structural and static coordination
• Identify what we need to coordinate
• Identify coordination policies (the how to)
• Exogenous model
• Based on the concept of Process coordination and
  Resource coordination

17

• In each step

18
Ongoing and Future Work (1/2)
minimize/maximize a function
19
Ongoing and Future Work (2/2)

• Process review
• To make it compositional
• Formalization
• By using a UML View model
• By using Queued Petri Nets
• Application to Industrial Process Planning
• To maximize/minimize the resource utilization
• Case Study
• Micron Italia

20
Combining the pieces together
drives
drives
1
Software Architecture
Software Architecture
2
4
Requirement Engineering
Design Coding
Coordination
Coordination
3
Coordination
validates
SEA group, Computer Science Dep., University of
LAquila, Italy
G. Papadopoulos University of Cyprus
21
Henry Muccini, Ph.D.Assistant ProfessorUniversit
y of LAquila Italymuccini_at_di.univaq.itwww.He
nryMuccini.com