1 / 35

Simon Perathoner 1 , Ernesto Wandeler 1 , Lothar Thiele 1

Influence of different system abstractions on the performance analysis of distributed real-time systems. Artist2 Meeting (Cluster on Execution platforms) Linköping, 14. May 2007. Simon Perathoner 1 , Ernesto Wandeler 1 , Lothar Thiele 1

zeus-alston
Download Presentation

Simon Perathoner 1 , Ernesto Wandeler 1 , Lothar Thiele 1

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Influence of different system abstractions on the performance analysis of distributed real-time systems Artist2 Meeting (Cluster on Execution platforms) Linköping, 14. May 2007 Simon Perathoner1, Ernesto Wandeler 1, Lothar Thiele 1 Arne Hamann 2, Simon Schliecker 2, Rafik Henia 2, Razvan Racu 2, Rolf Ernst 2 Michael González Harbour 3 3 Universidad de Cantabria 1 ETH Zürich 2 TU Braunschweig

  2. Outline • Motivation • Abstractions • Benchmarks • Conclusions

  3. System level performance analysis Max. CPU load? Max. end-to-end delay? Max. Bufferspace?

  4. Abstraction 1 Abstraction 2 Abstraction 3 Analysis method 1 TDMA GPC GPC GSC GPC Analysis method 2 Analysis method 3 GPC GPC Formal analysis methods Distributed system Abstraction Performance values Analysis

  5. Motivating questions • What is the influence of the different models onthe analysis accuracy ? • Does abstraction matter ? • Which abstraction is best suited for a given system ? Evaluation and comparison of abstractions is needed !

  6. How can we compare different abstractions ? modular performance metrics modeling effort correctness scope scalability usability analysis time accuracy exact efficientimplementation simple compositional holistic stepwise refinement tool support Method 1 learning curve easy to use Method 2

  7. B C A D What makes a direct comparison difficult? • Many aspects can not be quantified • Models cover different scenarios:

  8. Intention Compare models and methods that analyze the timing properties of distributed systems: • SymTA/S [Richter et al.] • MPA-RTC [Thiele et al.] • MAST [González Harbour et al.] • Timed automata based analysis [Yi et al.] • …

  9. MAST Approach • Leiden Workshop on Distributed Embedded Systems: http://www.tik.ee.ethz.ch/~leiden05/ • Define a set of benchmark examples that cover common area • Define benchmark examples that show the power of each method TA SymTA/S MPA

  10. Expected (long term) results • Understand the modeling power of different methods • Understand the relation between models and analysis accuracy • Improve methods by combining ideas and abstractions

  11. Contributions • We define a set of benchmark systems aimed at the evaluation of performance analysis techniques • We apply different analysis methods to the benchmark systems and compare the results obtained in terms of accuracy and analysis times • We point out several analysis difficulties and investigate the causes for deviating results

  12. Outline • Motivation • Abstractions • Benchmarks • Conclusions

  13. Abstraction 1 - Holistic scheduling Basic concept: extend concepts of classical scheduling theory to distributed systems Holistic scheduling FP + datadependencies[Yen et al.]1995 FP CPUs + TDMA bus[Tindell et al.]1994 FP + control dependencies[Pop et al.]2000 Mixed TT/ETsystems[Pop et al.]2002 EDF offset based [González et al.]2003 CAN[Tindell et al.]1995 . . .

  14. Holistic scheduling – MAST tool MAST - The Modeling and Analysis Suite for Real-Time Applications [González Harbour et al.] SystemTXT ResultsTXT

  15. Abstrction 2 – The SymTA/S approach Basic concept: Application of classical scheduling techniques at resource level and propagation of results to next component Problem: The local analysis techniques require the input event streams to fit given standard event models EMIF / EAF Solution: Use appropriate interfaces: EMIFs & EAFs

  16. SymTA/S – Tool

  17. Abstraction 3 – MPA-RTC availability events t t Service model Load model bu events service au bl al D D Arrival curves Service curves

  18. Abstraction 3 – MPA-RTC [bl, bu] [al, au] [al’, au’] GPC [bl’, bu’]

  19. Abstraction 4 - TA based performance analysis Verification of performance properties by model checking (UPPAAL) Exactperformance values periodic stream fixed priorityscheduling

  20. Outline • Motivation • Abstractions • Benchmarks • Conclusions

  21. Benchmarks • Pay burst only once • Complex activation pattern • Variable feedback • Cyclic dependencies • AND/OR task activation • Intra-context information • Workload correlation • Data dependencies

  22. Benchmark 1 – Complex activation pattern WCRT T3 ? (FP) periodic P = 60 C = 35 periodic P = 5 (FP) C = 2 C = 4 periodic C = 12 period

  23. Benchmark 1 – Analysis results

  24. Benchmark 1 – Result interpretation PI3 = 65 ms

  25. Benchmark 1 – Worst case Delay I2-O2

  26. Benchmark 2 – Variable feedback Exec. time C (FP) periodic P = 100 (FP) periodic P = 5 C = 2 C = 1 Max delay ? C = 2

  27. Benchmark 2 – Analysis results

  28. Benchmark 3 – Cyclic dependencies jitter (FP) periodic with burst P = 10 C = 4 C = 1 Max. delay ? C = 4

  29. Benchmark 3 – Analysis results Scenario 1: priority T1 = highpriority T3 = low

  30. Benchmark 3 – Analysis results Scenario 2: priority T1 = low priority T3 = high

  31. Analysis times [s]

  32. Outline • Motivation • Abstractions • Benchmarks • Conclusions

  33. Discussion • Approximation of complex event streams with standard event models can lead to poor performance predictions at local level • Holistic approaches better in the presence of correlations among task activations (e.g. data dependencies) • Cyclic dependencies represent a serious pitfall for the accuracy of compositional analysis methods • Holistic methods less appropriate for timing properties referred to the actual release time of an event within a large jitter interval

  34. Conclusions • The analysis accuracy and the analysis time depend highly on the specific system characteristics • None of the analysis methods performed best in all benchmarks • The analysis results of the different approaches are remarkable different even for apparently basic systems • The choice of an appropriate analysis abstraction matters • The problem to provide accurate performance predictions for general systems is still far from solved

  35. Thank you! Simon Perathoner perathoner@tik.ee.ethz.ch

More Related