A workflow-inspired, modular and robust approach to experiments in distributed systems

Abstract : Experimentation in large-scale distributed systems research is very challenging due to the size and complexity of modern systems and applications spanning domains of high performance computing, P2P networks, cloud computing, etc. Some obstacles that each researcher must face are: the difficulty of properly structuring experiments due to their complexity, the inflexibility of existing methodologies and tools and the scalability problems resulting from the size of studied systems. In this paper, we propose a novel method of representing and executing experiments that solves these problems. To this end, we present an interdisciplinary approach to the control of large-scale experiments in distributed systems research that draws its foundations from workflow management and scientific workflows. This workflow-inspired approach distinguishes itself by its representation of experiments, modular architecture and robust error handling. We show how the aforementioned problems are solved by our approach in an exemplary performance study of an HTTP server.
Liste complète des métadonnées

https://hal.inria.fr/hal-00909347
Contributor : Lucas Nussbaum <>
Submitted on : Monday, June 23, 2014 - 5:01:29 PM
Last modification on : Tuesday, December 18, 2018 - 4:26:02 PM
Document(s) archivé(s) le : Tuesday, September 23, 2014 - 12:06:00 PM

Files

Identifiers

  • HAL Id : hal-00909347, version 5

Citation

Tomasz Buchert, Lucas Nussbaum, Jens Gustedt. A workflow-inspired, modular and robust approach to experiments in distributed systems. CCGRID - 14th IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing, May 2014, Chicago, United States. ⟨hal-00909347v5⟩

Share

Metrics

Record views

643

Files downloads

422