ECOO

1
ECOO
Environments and COOperation
2
ECOO Environments and COOperation
e_work Cooperative Work Internet
Co-design Co-engineering
Coordination of a Virtual Team
Process modeling, Workflow Awareness Collaborative
Tools
3
ECOO Members
Responsable Claude Godart, Prof. UHP, ESSTIN
Phd students Karim Baina, Julia Bitcheva
Christophe Bouthier Abdelmajid
Bouazza Daniela Grigori Post docs
Samir Tata (Aalborg, Dan)
Olivier Zendra (McGill, Can) Engineer Saâd El
Hadri Marc Patten
Permanents Khalid Benali, MC Nancy 2 Nacer
Boudjlida, Prof UHP Gérôme Canals, MC Nancy
2 François Charoy, MC Nancy 2, CR INRIA (sept 00
-- sept 02) Jean-Claude Derniame, Prof
ENSEM Jacques Lonchamp, Prof. Nancy 2 Pascal
Molli, MC UHP, CR INRIA (sept 00 -- sept
02) Olivier Perrin, MC Nancy 2 Hala Skaf-Molli ,
MC UHP
4
Guidelines

• Develop tools
• that people want
• easy to deploy and debug (amateur programming)

5
Industrial relationships

• AEE (Car manufacturers and equipment providers),
  E! EAST
• Interoperability of processes
• CNET (France-Télécom), CRAI
• Virtual teams for architects
• Hitachi Japan
• Flexible workflow
• E! KVM (Kappa, HTTP Software)
• ...

6
Academic relationships

• ISA, MACSI, MAIA, ORPAILLEUR, TRIO
• ADP, EIFFEL ...
• Working group on virtual teams with UWS, UNSW,
  Macquire U à Sydney
• HPLab Palo Alto
• ...

7
Virtual Teams

• Team
• A group of people who interact through
  interdependent tasks guided by a common goal
• Virtual
• Distributed across the time, space and the
  organizations
• Linked by webs of communications technologies

8
Virtual Teams

• Why virtual teams ?
• Sharing
• Quickly gathered core competencies of different
  organizations
• Work is achieved in a competitive time
• High-productivity, mobility
• Sharing

9
Virtual Teams

• Providers host virtual teams
• BSCW Reference of CSCW community for shared
  workspace, a commercial product now
• SourceForge 10000 open source projects, 70 000
  virtual members
• HotOffice, TeamCenter, Egroups, Instant!TEAMROOM
  Commercial products

10
Virtual Team services

• Shared data
• Object and dependencies, Concurrency, versioning,
  annotation, Access Control, Notification
• Communication Services
• Synchronous video-conference, application
  sharing, collaborative tools, shared whiteboard,
  chat
• Asynchronous Mailing list, web pages, forums,
  persistent notification

11
Virtual Team Services

• Coordination Tasks management
• Manual Project Management
• Enacted To_Do_List
• Defined in a process and enacted Workflows
• Awareness pertinent
• Activity awareness, Availability awareness,
  Process awareness.

12
ECOO Research themes

• Shared data consistency
• Concurrent access to shared data
• Asynchronous vs. synchronous updates
• Coordination
• Implicit coordination (informal)
• Explicit coordination (formal)
• Collaboration

13
Problems

• Shared data consistency
• Concurrent access to shared data
• Asynchronous vs. synchronous updates
• Coordination
• Implicit coordination (informal)
• Explicit coordination (formal)
• Collaboration

14
Problems

• Shared data consistency
• Concurrent access to shared data
• Asynchronous vs. synchronous updates
• Coordination
• Implicit coordination (informal)
• Explicit coordination (formal)
• Collaboration

15
Shared Data in Virtual Teams
Private Workspace
Private Workspace
Private Workspace
Shared Workspace
Private Workspace
Private Workspace
Private Workspace
Private Workspace
16
Shared Data and Copy-Modify-Merge
Private Workspace
Private Workspace
Private Workspace
Shared Workspace
Private Workspace
Private Workspace
Private Workspace
Private Workspace
17
Shared Data and Consistency
a
b
C
18
Data Consistency

• Transactional approach
• Encapsulate activities in transactions.
• The system will take in charge the problem of
  concurrent access

19
Data Consistency
Private Workspace
Private Workspace
Private Workspace
Shared Workspace
Private Workspace
Private Workspace
Private Workspace
Private Workspace
20
Which transaction model ?

• Long term
• Uncertain
• Non isolated
• ...

21
Cooperation Patterns

• Client/Server
• NO concurrent modifications on shared data

• Redactor/reviewer
• Cross reading of shared data

• Cooperative Writing
• Concurrent modifications

22
Data Consistency

• Results
• Coo-serializability new correction criterion
  for cooperative executions
• Coo-transactions new advanced transaction model
  based on coo-serlializability
• COO prototype a cooperative development
  environment based on the coo-transactions.
• Canals98, Information sciences, 110, p 297--302

23
Problems

• Shared data consistency
• Concurrent access to shared data
• Asynchronous vs. synchronous updates
• Coordination
• Implicit coordination (informal)
• Explicit coordination (formal)
• Collaboration

24
Project life cycle
Divergence concurrent modifications on the
different copies of the object. Convergence
integration of parallel modifications ...
Divergence

time

User1

User2

synchro. Points
User3
25
Problem

• Currently, divergence and convergence phases are
  disconnected.
• How to increase convergence efficiency by
  establishing a continuity between asynchronous
  and synchronous phases ?

26
Basculement vers une session de Travail Couplé
User 1
User 2
User 3
User 4
27
Basculement vers une session de Travail Couplé
Conflit résolu
User 1
User 2
User 3
User 4
28
Fin de la session de Travail Couplé
User 1
User 2
User 3
User 4
29
General approach
Late validation of Operation (like CVS)
Immediate validation of operation (like Ellis)
Reconciliation stage
WS1
WS1
WS2
WS2
WS3
WS3
30
Conclusion

• Expected results
• Reconciliation of the tightly coupled way of
  working (like in real time groupware, no
  divergence) and the loosely coupled approach
  (like in CM, Divergence allowed
  Copy/modify/merge)
• Scientific contribution
• A common model for late validation and immediate
  validation scheme
• Bouazza99, CSCW Workshop on real time edition

31
Problems

• Shared data consistency
• Concurrent access to shared data
• Coordination
• Explicit coordination (formal)
• Implicit coordination (informal)
• Collaboration

32
Formal Coordination

• Cooperative processes
• Workflow is used in several domains to coordinate
  works
• Example coordinate tasks in office automation
• Use it to coordinate virtual teams !
• Coopetitive processes

33
Formal Coordination

• Cooperative processes
• Workflow is used in several domains to coordinate
  works
• Example coordinate tasks in office automation
• Use it to coordinate virtual teams !
• Coopetitive processes

34
Workflow Management System

• Two components
• Process definition (declarative or/and visual)
• Define tasks ...
• Who must accomplish the task
• Define transitions between tasks.
• Process execution engine
• Execution environment
• Users tasks appear on their desktop...

35
Example Loan Request Handling

• Activities
• Collect credit information
• Assess risk (amount lt 10 000 FF)
• Request approval (amount gt 10 000 FF or high
  risk)
• Accept credit
• Reject credit

36
Control flow of activities
37
My activities
38
Problem

• Actual workflow systems are rigid (competitive
  activities)
• End-gtbegin dependency
• Execute in isolation
• Add flexibility to workflow systems

39
Cooperative Workflow

• Two methods
• Change the execution engine
• Keep the model simple (minimal modifications)
• Accept cooperative executions (break isolation)
• Change the process model
• Add cooperative operators to the model

40
Competitive vs. Cooperative Process Interpretation
Process Model
Competitive Interpretation
Cooperative Interpretation
41
Principle

• 1. Activities executes as COO-transactions
• 2. An activity can anticipate its execution
• start its execution with intermediate results
• even if all its conditions are not fulfilled

42
Activities as COO-transactions
C

A



B

Cooperation supported by cooperative transactions



Cooperation supported by anticipating activities

43
Anticipated activities
C

A



B

Cooperation supported by cooperative transactions



Cooperation supported by anticipating activities

grigori01a, DEXA conference
44
Example

• Activity goes to the anticipating state
• User choose when they want to start them
• At some time they must go to the executing state
• From the outside the execution remains the same

45
Cooperative workflow
C

A



B

Cooperation supported by cooperative transactions



Cooperation supported by anticipating activities

grigori01b, CODAS conference
46
Condition for anticipation

• When an activity can become ready to anticipate ?
• Free anticipation
• All activities may anticipate
• Possible but difficult to converge to a
  consistent state
• High risk of lost or unuseful work
• Control flow dictated anticipation
• Start-start dependency with preceding activities
• Reduce the risk of lost work
• Data flow dictated anticipation (require explicit
  data flow definition)
• The mandatory input must be available
• Guarantee the availability of all the ressources
  (may be not in the right state

47
Anticipation and Coo transactions

• Anticipation allows flexible execution but
  increased the risk of inconsistencies
• Coo transactions allows flexible data management
• Combination of both will support a form of
  coordinated cooperation

48
Combining anticipation and the Coo protocol

• The Coo protocol generates dynamic dependencies
• Dynamic dependencies must not contradict static
  dependencies (of the process)
• gt Constraints on intermediate results published
  by anticipating activities
• Anticipating activities may publish only if they
  are certain to go in executing state at some time
• Intermediate results of anticipating activities
  cannot be read by executing activities (risk of
  cycle)

49
Anticipation and Coo transaction

• Data and workflow management are integrated
• C/S between successive activities
• CW and R/R between parallel activities (and even
  processes)
• Users have some initiative but remains under
  control
• The design of the process remains simple
• Designer may rely on its flexibility

50
Cooperative Workflow

• Two methods
• Change the execution engine
• Keep the model simple (minimal modification)
• Take in consideration cooperative executions
• Change the process model
• Add cooperative operators to the model

51
Change the Process Model

• Cooperation is defined and controlled explicitly
• 3 cooperation patterns
• Client server
• Redactor reviewer
• Cooperative writing
• Users can negotiate their cooperation strategy

godart99, RIDE Virtual Enterprises
52
Cooperation Patterns

• Client/Server
• NO concurrent modifications on shared data

• Redactor/reviewer
• Cross reading of shared data

• Cooperative Writing
• Concurrent modifications

53
Example Cooperative Write
A
Edit
Shared workspace
B
Edit
Cooperative Operator
54
Formal Coordination

• Approach based on workflow (explicit
  coordination)
• Workflow is used in several domains to coordinate
  works
• Example coordinate tasks in office automation
• Use it to coordinate virtual teams !
• Coopetitive processes

55
Problem

• Two enterprises cooperate on one project but
  compete on another
• They need to cooperate, but they do not want to
  put on the table all their know_how
• How to outsource a process ? How to answer to an
  invitation to tender ?

56
Coopetitive processes
Modèle de procédé interentreprises
a
b
c
d
Généralisation/spécialisation
a
c
b
d
GP
Gestionnaire de procédé
57
Approach ?

• Process bus
• how to abstract a process ?
• More than a programming object (duration,
  uncertainty )

58
Problems

• Shared data consistency
• Concurrent access to shared data
• Coordination
• Explicit coordination (formal)
• Implicit coordination (informal)
• Collaboration

59
Informal Coordination

• Gutwin, Roseman and Greenberg in CSCW96
• "Workspace awareness reduces the effort needed
  to coordinate tasks and resources, help people to
  move between individual and shared activities,
  provides a context in which interpret utterance,
  and allows anticipation of others actions

60
Informal Coordination

• Coordination based on group awareness
• Keep team members informed about the project
  activity..
• So they will communicate...
• And auto-coordinate.
• Essential problem
• Providing pertinent information.
• Which information ? when ? To whom ?

61
Informal Coordination

• A pertinent source of information
• The measurement group divergence.
• Group divergence
• Measure the quantity of disorder between the
  different copies of shared data...

62
Divergence
Divergence concurrent modifications on the
different copies of the object. Convergence
integration of parallele modifications ...
Divergence

time

User1

User2

synchro. Points
User3
63
Divergence/convergence

• Divergence phase Monitor...
• Convergence phase Accelerate...
• Convergence points COO-SR

64
Divergence

• How to measure divergence?
• How to propagate it ? According to which model ?
• How to visualize it in a real time ?

65
How to Measure Divergence?

• Configuration management approach
• Copy/modify/merge calculate the difference
  between the different copies of an object
  flexible merge, diff
• Transactional approach
• Number of dirty reads ?-serializability
• Operational transformation
• Number of conflict operations Ellisgibbs,
  SOCT4.

66
Divergence Propagation

• Divergence is the quantity of disorder in objects
  network
• It propagates through dependencies between
  objects ...

67
Divergence Visualization

• User must be able to monitor divergence
  variation like CPU load or network activity.
• Peripheral, scalable, customizable...

68
Divergence Visualization
69
Divergence Visualization
70
Divergence Visualization
71
Experimentation

• Prototype
• MOTU prototype is under realization in an
  advanced step..
• motu.sourceforge.net
• Hypothesis
• If users visualize divergence in a real time so
  they will communicate to auto-coordinate.
• Experience
• Clinical studies on virtual teams disposed on
  information about divergence.
• This is the objective of the project COCAO with
  CNET.

72
Conclusion on Informal Coordination

• Divergence is a pertinent source of notification
• A peripheral visualization allows user to
  monitor divergence like CPU load or network
  activity..
• Auto-coordination

73
Research themes

• Shared data consistency
• Concurrent access to shared data
• Asynchronous vs synchronous updates
• Coordination
• Implicit coordination (informal)
• Explicit coordination (formal)
• Collaboration

74
Collaboration

• Decision support
• Negotiation
• Collective resolution of questions
• Argumentation
• ...
• http//www.loria.fr/equipes/ecoo/dots/

75
General Conclusion

• Data consistency
• Coo-transactions
• Immediate vs. differed updates
• Coordination two approaches
• Formal coordination
• Cooperative workflow
• Coopetitive processes
• Informal coordination
• Auto-coordination based on group awareness
• Divergence metrics

76
motu.sourceforge.net