A Reference Architecture for Component-Based Self-Adaptive Software Systems. Lorena Castañeda Bueno Advisor: Gabriel Tamura

Transcription

1 A Reference Architecture for Component-Based Self-Adaptive Software Systems Lorena Castañeda Bueno Advisor: Gabriel Tamura

2 Motivation Today s software applications face unpredictable context changes, including business goals changes in organizations Software systems must fulfill these changes in requirements in a very short time, and especially at runtime. Software adaptation is expensive and takesalotoftimeandeffort. 2

3 Motivation Lets see an example Mobile Banking Service Manage Accounts Payments Transfers THIS REQUIRES SYSTEMS THAT CAN BE MORE new Secureand Fast SELF-MANAGEABLE Security protocols Cryptography Too much time! 3

4 Motivation Software engineering for Self-Adaptive Systems communityhave made significant efforts to propose self-adaptive systems, however: Not general enough to fit the different systems to build self-adaptive software Adaptation mechanisms merged with the target system. Feedback loops are not explicit in most of these proposals. 4

5 Presentation outline 1. Background 2. TheProblem 3. Solution and Contributions 4. Evaluation 5. Conclusions and Future Work 5

6 Background Knowledge Areas: 1. Autonomic Computing and Software Adaptation 2. Self-adaptive Software System Approaches 3. Component-Based Software Engineering 4. Software Architecture, Reference Modelsand Reference Architectures 6

7 1. AutonomicComputing and Software Adaptation Self-managed systems: The capability to run routine tasks and maintenance without the intervention of the system administrator. Adaptation refers to the capability of changing the software structure and behavior to meet the changing business requirements. Feedback is the process to return an output of the system to check if the control objectives are being achieved 7

8 Control loop Control loop is the feedback mechanism to control the dynamic behavior of the system, supported in four activities: Collect Relevant data from the system and environment Analyze the collected data Decide how to adapt in order to reach a desirable state Act realize in the system the decisions made by the control Software engineering for self-adaptive systems: A research roadmap B. H. Cheng et al. 8

9 2. Self-adaptive software system approaches 1. IBM s autonomic computing - ACRA 2. Kramer and Magee, three layer reference architecture 3. Garlan, the Rainbow Framework based on reusable architecture and external adaptation 4. Parra s Cappucino architecture 5. Solomon s autonomic system 6. DYNAMICO reference model 9

10 Autonomic computing reference architecture (ACRA) - IBM Level 5 Level 4 Level 3 Level 2 Level 1 An Architectural blueprint for autonomic computing, IBM,

11 LEVEL 3 - Autonomic manager It is the component that implements the control loop: 1. Monitor 2. Analyzer 3. Planner 4. Executer 5. These four elements communicate to and collaborate with each other and exchange appropriate Knowledge and information. 11

12 LEVEL 2 Touchpoints Components that expose the state and management operations for a managed resource To sense changes (properties and events) Managed Resources To effect changes (operations) Target System Data sources Other systems 12

13 DYNAMICO: A Reference Model for Governing Control Objectives and Context Relevance in Self-Adaptive Software Systems Control objectives Adaptation Control Context control Villegas, N.M., Tamura, G., Müller, H.A., Duchien, L., Casallas, R, LNCS Vol.7475,

14 3. Component-Based software engineering SOFTWARE COMPONENT Software elements Independently developed and deployed Hides all the complexity of the software implementation Components reuse Permits easy replacement Component-basedsoftware engineering, George Heinemanand William Councill,

15 4. Software Architecture, Reference Models and Reference Architectures An architecture in software defines the elements that compose the system and the relationshipbetween those elements. Decomposition of a well known type of problem Mappingbetweena Reference Model, ArchitecturalPatternsand thedata flowamongthem A set of elements in component types and the relationships among them Software Architecture In Practice, Len Bass and Paul Clements and Rick Kazman,

16 2. THE PROBLEM 16

17 The Problem Changing context and software adaptation Different systems, data types configurations Self-Adaptation merged with the Target System Feedbackloopsare not explicit To build self-adaptive systems with heterogeneous available systems The necessity to manage adaptation mechanism Measure the output of the system and evaluate adaptation 17

18 3. PROPOSED SOLUTION 18

19 Our Reference Architecture PROJECT GOALS: 1. To design and implement a componentbased reference architecture for component-based self-adaptive software systems. Separation of concerns Distribution Extensibility 2. To evaluate our proposal In an specific case of study 19

20 1. Control Feedback loop 2. DYNAMICO Our high level goal is to provide software engineers with a guide to design software architectures for self-adaptive systems + + Data Flow + 1. Pipes and Filters 2. Event-Based 3. Blackboard An Introduction to software architecture, David Garlan and Mary Shaw,

21 Architecture Data Flow Target System Adaptation mechanism THE INFORMATION IS CONTEXTRELATED 21

22 Architecture Data Flow CONTEXT METAMODEL byvillegas et al. Our reference architecture context metamodel Context-Driven Adaptive Monitoring for Supporting SOA Governance, Villegas and Müller,

23 Architecture Data Flow CONTEXT METAMODEL subset Diagnosis Context Event information Adaptation Actions Context Data Commands Context-Driven Adaptive Monitoring for Supporting SOA Governance, Villegas and Müller,

24 Our Architecture Data Flow Proposal 24

25 Architecture Data Flow A SLO describes that for a given day, the Maximum of Transactions per minute allowed is 100, if more then adaptation is required In thelastminute 115 (Tx/min) CurrentTx/min is115. Desired Tx/min for today is > 100 -> Adaptationisrequired Registration request (Tx/min) 25

26 + + Data Flow + 26

27 Our Reference Architecture for Component-Based Self-Adaptive Software Systems Implementation Adaptation Mechanism Human Interaction Context Interaction 27 27

28 Our Reference Architecture for Component-Based Self-Adaptive Software Systems Implementation 115 Registration request in the last minute Current : 115 Desired: 100 Adaptation Analyzing Policies SLO: Max 100 Monitoring policies Registration request Planning policies Adaptation Actions Commands 28

29 General Representation An Interface to expose services An Abstract class to define basic behavior A Specific implementation 29

30 Java Implementation 30 30

31 4.EVALUATION 31

32 Evaluation Evaluate a concrete architecture derived from our Reference Architecture A Case of Study Explore its extensibility to support dynamic adaptation of SCA Software Applications Explore its distributionexecuting the components separately 32

33 Reliable Videoconference Service (RVCS) Video-conference client Registers and Attends Establishment of Service Conditions (QoS) Video-Conference System 33

34 A concrete SCA Architecture derivation from our Reference Architectura Each component is described in its own composite One composite file to group all components Service Component Architecture, Assembly Model Specification. Specification Version 1.0, Beisiegel, M. and Blohm, H. and Booz, D. and Edwards, M. and Hurley, O.,

35 Testing Deployment Configuration Not implemented 35

36 Test Cases and Results 36

37 Test Results Amountof Requests Evaluation criteria SLO Fulfillment Adaptation Requirement 37

38 Test 2 Results 38

39 Test 2 Results 39

40 Test 3 Results 40

41 Evaluation Results A Case of Study Extensibility: SCA Software Applications Distribution: Separating MAPE-K from Sensor and Effector 41

42 5. CONCLUSIONS AND FUTURE WORK 42

43 Conclusions(I) Our reference architecture provides a generic framework to implement adaptation mechanisms for software systems. Separation of concerns enabling the evaluation, maintenance and analysis of the adaptation process in a selfadaptive system. Data flow modeling using an extensive context metamodel 43

44 Conclusions(II) Based on components, this reference architecture also provides component reuse, the ability to be extensible and interoperable, which can help software engineers to specialize, extendand growthe functionalities of each component, according to the necessities of the specific target system. 44

45 Conclusions(III) As an SCA-compliant implementation, our reference architecture fulfills the goal of being extensible Each SCA component of the reference architecture is defined in its own composite file in order to fulfill the distribution capability 45

46 Future Work Security between the Target System and the MAPE-K AMC development to assist humans in the definition of SLOs. Include V&V modeling in the Reference Architecture V&V after adaptation Planner and Executer extensions 46

47 References [1] Len Bass, Paul Clements, and Rick Kazman. Software Architecture In Practice. Addison-Wesley, [2] M. Beisiegel, H. Blohm, D. Booz, M. Edwards, and O. Hurley. Service Component Architecture, Assembly Model Specication. Specication Version [3] B. H. Cheng, R. Lemos, H. Giese, P. Inverardi, J. Magee, J. Andersson, B. Becker, N. Bencomo, Y. Brun, B. Cukic, G. MarzoSerugendo, S. Dustdar, A. Finkelstein, C. Gacek, K. Geihs, V. Grassi, G. Karsai, H. M. Kienle, J. Kramer, M. Litoiu, S. Malek, R. Mirandola, H. A. M uller, S. Park, M. Shaw, M. Tichy, M. Tivoli, D. Weyns,, and J. Whittle. Software engineering for self-adaptive systems: A research roadmap. Software engineering for self-adaptive systems, LNCS 5525:1 { 26, [4] Microsoft Corp. Introducing Windows Communication Foundation in.net Framework 4. (Last visited: Jan/2012). [5] R. Dawson, R. Desmarais, H.M. Kienle, and H.A M uller. Monitoring in Adaptive Systems Using Reection. ACM/IEEE, SEAMS 2008:81 { 88, [6] A.G. Ganekand T.A. Corbi. The dawning of the autonomic computing era. IBM Systems journal, 42(1):5{18, [7] David Garlan, Shang-Wen Chen, An-Cheng Huang, Bradley Schmerl, and Peter Steekiste. Rainbow: Architecture-based self-adaptation with reusable infrastructure. IEEE Computer, 37(10):46{54,

48 References [8] David Garlan and Mary Shaw. An Introduction to software architecture. School of computer science. Carnegie Mellon University, [9] H. Giese, Y. Brun, J. D. M. Serugendo, C. Gacek, H. Kienle, H. M uller, M. Pezze,, and M. Shaw. Engineering self-adaptive systems through feedback loops. Springer- Verlag, LNCS 5525:47 { 69, [10] George Heineman and William Councill. Component-based software engineering. Addisonwesley, [11] J. L. Hellerstein, Y. Diao, S. Parekh,, and D. M. Tilbury. Feedback Control of Computing Systems. John Wiley and Sons, [12] IBM. An Architectural blueprint for autonomic computing. Autonomic computing, [13] Oracle Inc. Java EE at a Glance. index.html (Last visited: Jan/2012). [14] J.O. Kephart and D.M. Chess. The vision of autonomic computing. IEEE Computer, 36 No.1:41 { 50, [15] J Kramer and J Magee. Self-managed Systems: an Architectural Challenge. IEEE Computer, pages 259 { 268,

49 References [16] Philip K McKinley, SeyedMasoudSadjadi, and Betty H.C. Cheng. Composing Adaptive Software. IEEE Computer, July [17] H. Müller, M. Pezze,, and M. Shaw. Visibility of control in adaptive systems. Proceedings of the 2nd international workshop on Ultra-large-scale software-intensive systems, ULSSIS 2008:23 {26, [18] H. A. Müller, H. M. Kienle, and U. Stege. Autonomic computing: Now you see it, now you dontdesignand evolution of autonomic software systems. Lecture Notes in Computer Science, 5413:32 { 54, [19] OASIS Organization. Service Component Architecture (SCA) Specication. ttp:// (Last visited: Jan/2012). [20] Carlos Parra, Xavier Blanc, and Laurence Duchien. Context Awareness for Dynamic Service-Oriented Product Lines. In John McGregor and Dirk Muthig, editors, 13th International Software Product Line Conference, volume 1, pages 131{140, San Francisco, Etats-Unis, August Acceptance rate: 30/83 (36%). Rank (CORE) : B. [21] BogdanSolomon, Dan Ionescu, Marin Litoiu, and MirceaMihaescu. A real-time adaptive control of autonomic computing environments. CASCON, pages 124 { 136, [22] Thomas Stahl and Markus Vlter. Model-Driven Software Development. JhonWiley and Sons, Ltd,

50 References [23] Clemens Szyperski. Component Software. Beyond Object-Oriented Programming. Addisonwesley, [24] G. Tamura, C. Demarey, R. Casallas, and L. Duchien. QoS-CARE: A Framework for selfreliable QoSContract Preservation through Self-Reconguration. Preprint submitted to JSS Special Issue State of the Art in Self-Adaptive Systems, [25] Gabriel Tamura. QoS-CARE: A Reliable System for Preserving QoS Contracts through Dy-namic Reconguration. Phdthesis, University of Lille 1 -Science and Technology and Universidad de Los Andes, May [26] Gabriel Tamura, Rubby Casallas, Anthony Cleve, and Laurence Duchien. QoS contractaware recongurationof component architectures using e-graphs. In Proceedings of the 7 th international conference on Formal Aspects of Component Software, FACS'10, pages 34{52, Berlin, Heidelberg, Springer-Verlag. [27] Vijay Tewariand Milan Milenkovic. Standards for Autonomic Computing. Intel technology journal, 10 No.4:275 { 284, [28] D. Truex, R. Baskerville, and H Klein. Growing Systems in Emergent Organizations. Communicationsof the ACM, 42 No. 8:117 { 123,

51 References [29] N.M. Villegas and H.A M uller. Context-Driven Adaptive Monitoring for Supporting SOA Governance. Proceedings of the 4th International Workshop on a Research Agenda for Mainte-nanceand Evolution of Service-Oriented Systems (MESOA 2010), Carnegie Mellon University Software Engineering Institute, [30] NorhaM. Villegas, Gabriel Tamura, HausiA. M uller, Laurence Duchien, and Rubby Casallas.DYNAMICO: A Reference Model for Governing Control Objectives and Context Relevance in Self-Adaptive Software Systems. In Software Engineering for Self-Adaptive Systems 2, volume 7475 of LNCS, pages 282 { 310. Springer,

52 52