Presentation is loading. Please wait.

Presentation is loading. Please wait.

WIR FORSCHEN FÜR SIE The Palladio Component Model (PCM) for Performance and Reliability Prediction of Component-based Software Architectures Franz Brosch.

Similar presentations


Presentation on theme: "WIR FORSCHEN FÜR SIE The Palladio Component Model (PCM) for Performance and Reliability Prediction of Component-based Software Architectures Franz Brosch."— Presentation transcript:

1 WIR FORSCHEN FÜR SIE The Palladio Component Model (PCM) for Performance and Reliability Prediction of Component-based Software Architectures Franz Brosch 28/11/2011

2 1.Foundations 2.PCM overview 3.Performance & reliability prediction 4.PCM applications & further research Outline 2

3 Software System Quality 3 Foundations Overview Prediction Applications Quality Requirements Actual Quality Product (Software System) Requirements Specification Software Development Process

4 Design-time Modeling & Prediction 4 Architectural model of a component-based software system Prediction of expected quality  Performance  Reliability  … Foundations Overview Prediction Applications

5 AUTOMATION SUPPORT Architectural Design Loop 5 Predicted Quality Metrics Evaluation Feedback Architectural Model Estimation, Benchmark Annotated Architectural Model Quality models (e.g. queueing networks for Performance) Model Transfor- mation Analysis / Simulation Foundations Overview Prediction Applications

6 The Palladio Approach 6 A Component Model Multiple Analysis Methods A Development Process Foundations Overview Prediction Applications

7 Component Quality Influences 7 User behaviour Hardware External Services Code Foundations Overview Prediction Applications

8 PCM Developer Roles 8 Foundations Overview Prediction Applications

9 Models and Analyses 9 Domain Expert System Deployer Software Architect Component Developer Transformation Java Code Skeletons Completion + Compilation Transformation Queueing Network Simulation Transformation Performance Prototype Execution + Measurement Part of PCM Instance Foundations Overview Prediction Applications Markov Chains Analysis

10 PCM Bench 10 Foundations Overview Prediction Applications

11  Comprehensive quality modeling approach  Improved support for CBSE process  Improved prediction accuracy Sophisticated usage model & usage propagation Stochastic Expression language for specification of arbitrary probability distributions and parametric dependencies  Support for trade-off analyses  Tool support available PCM Features 11 Foundations Overview Prediction Applications

12 Performance Prediction 12 > RDSEFF > RDSEFF Service Call CPU HDD Memory Resource demandsCommunication link latency & throughput Hardware resource processing speed System workload Foundations Overview Prediction Applications Resource utilization > Response time Resource utilization Response time Resp. time Throughput Return

13  Quality models Queueing network (evaluation through simulation)  Result metrics Response times (system level / component level) Throughput Hardware resource utilization  Features Developer roles, UML-like modeling, usage propagation Multi-user scenarios with concurrent behavioral specification Arbitrary input distributions / full result distributions Performance Prediction 13 Foundations Overview Prediction Applications

14 Reliability Prediction 14 > RDSEFF > RDSEFF Service Call Return CPU HDD Memory Software failures Communication link failures Hardware resource unavailability Foundations Overview Prediction Applications P(SUCCESS)

15  Quality model Set of absorbing discrete-time Markov Chains (DTMC) Evaluation through analysis  Result metrics Probability of successful execution of a usage scenario (= 1 – POFOD)  Features Developer Roles, UML-like modeling, usage propagation Accurate modeling of control and data flow Combined consideration of software reliability & hardware availability Reliability Prediction 15 Foundations Overview Prediction Applications

16 PCM Applications: Research 16 Foundations Overview Prediction Applications CoCoME Common Component Modelling Example Academic modelling contest SLA@SOI SLA management and service-oriented infrastructures Application of PCM to service-oriented systems Q-ImPrESS Integrated architecture-based quality modelling & prediction PCM provides conceptual and tooling input

17 PCM Applications: Industry 17 IBM zSeries Mainframe Design alternatives for storage virtualization SAP Service-based system Sizing of resources CAS CRM-Software Quality prediction for architecture evolution scenarios PTV Navigation software Web service response time evaluation Foundations Overview Prediction Applications

18 Further Research Directions 18  Architecture optimization  Architecture reengineering  Run-time prediction  Maintainability  Security  Component certification Foundations Overview Prediction Applications

19 Thank You 19 Thank you!


Download ppt "WIR FORSCHEN FÜR SIE The Palladio Component Model (PCM) for Performance and Reliability Prediction of Component-based Software Architectures Franz Brosch."

Similar presentations


Ads by Google