Running experiments on modern systems such as supercomputers, cloudinfrastructures or P2P networks became very complex, both technicallyand methodologically. It proved difficult to run experiments correctlyand understand obtained results, even with the background on theemployed technology and methods. Moreover, large-scale experimentssuffer from erroneous and the unpredictable behavior of underlyingsoftware and hardware, undermining the scientific principles ofexperimental computer science. This worrisome state of research onlarge-scale distributed systems calls for new approaches to design,run and interpret experiments.This work explores the use of control-flows (business processes) as amodel for representing the large-scale experiments in research ondistributed systems. We set out to find advantages, disadvantages andlimitations of this approach, and practical considerations for futureimplementers.We make 3 main contributions. First, we analyze the current state ofexperiment management tools, their limits and features to betterunderstand difficulties that lay ahead. We construct a generalframework to evaluate tools of this type. Second, we design andimplement an experiment management tool which is based on the model ofcontrol-flows. We show that this methodology can be implemented andused in practice to run challenging and large-scale experiments whileoffering a wide set of features, some of them missing in the previousapproaches. Finally, we analyze the use of provenance in computerscience, and in particular in experimental research on distributedsystems, and propose a provenance collection system that emerges fromthe control-flow model used as the representation of experiments. Thedesign is implemented and shown to collect provenance in efficient andautomatic way.Our results show that workflows are a viable model for the design andexecution of experiments in distributed systems research. With thesepositive conclusions in mind, we also sketch future researchdirections for improving our work.

EXPLOR_STEP=$WICRI_ROOT/Wicri/Lorraine/explor/InforLorV4/Data/Main/Exploration

HfdSelect -h $EXPLOR_STEP/biblio.hfd -nk 000092 | SxmlIndent | more

HfdSelect -h $EXPLOR_AREA/Data/Main/Exploration/biblio.hfd -nk 000092 | SxmlIndent | more

{{Explor lien
   |wiki=    Wicri/Lorraine
   |area=    InforLorV4
   |flux=    Main
   |étape=   Exploration
   |type=    RBID
   |clé=     Hal:tel-01273964
   |texte=   Managing large-scale, distributed systems research experiments with control-flows
}}