Elements: Software: Distributed Workflows for Cyberexperimentation (DEW)

Today’s testbed experimentation is ad-hoc, lengthy and manual. This makes it hard to reuse experiments and reproduce results. Sometimes even authors struggle to reproduce their prior work. 

Testbeds focus on experiment’s physical design, not the entire lifecycle. This poster describes our project, which t supports experiments though the entire lifecycle: through design, running, monitoring, storage, analysis, sharing and reuse stages. We do so by creating a new experiment representation, called DEW. DEW has three parts: scenario - a human-readable list of actors and actions as well as their dependencies, constraints - physical requirements for testbed nodes and topology on which the scenario will run and bindings - actual commands to run for each action and each dependency from the scenario. On top of this common experiment representation, we build other support services for experiment lifecycle, and we build translators that generate different topology and runtime scripts for different testbeds.