A demonstration of cross-domain workflow building on the DAFNI platform

DAFNI (Data and Analytics Facility for National Infrastructure) is a major UK national facility, based at the Science and Technology Facilities Council, which supports cross-disciplinary collaborative research in the infrastructure domain. Infrastructure is a vital factor in the development of modern societies. The ongoing maintenance, construction, and development of our infrastructure systems and networks is critical for the economic, social, and physical wellbeing of the nation, as well as building our resilience to potential disruptors such as climate change.Historically, computational models of infrastructure systems have tended to focus on isolated research domains, in contrast to their real-world counterparts which can impact one another as they are often physically proximate. Due to differences in areas such as programming standards, data, and data formats between research groups, the coupling of these models to account for their real-world interaction is a complex task. This can lead to duplication of effort and loss of knowledge, slowing down innovation efforts.The DAFNI platform was built as a solution to these challenges and forms a central resource to infrastructure researchers in the UK. The platform provides long-term storage of datasets and software models, as well as a record of the modelling activities undertaken, standing as a legacy beyond the lifetime of a project. The preservation of historical data and models is central to the DAFNI project.In addition to promoting important research work on infrastructure and providing a clear framework for model and data provenance, a major feature of the DAFNI platform is to allow researchers to combine independent models together to form workflows, supporting systems-of-systems modelling. Using containerisation techniques, researchers can upload and run their models in workflows while accessing a common data repository along with a selection of visualisation tools. DAFNI supports workflows which allow researchers to use the outputs of one model as the inputs for another, connecting models of different systems together.Providing a structured and abstracted way of running reproducible and sharable models, the workflow building and execution environment is easily accessible through the DAFNI web-app. The intuitive interface provides users with a drag and drop canvas to construct and run multi-step workflows, allowing for easily chaining models, and removing the need for any command line knowledge in comparison to other computational services.A demonstration of the building and running of a workflow on DAFNI will be presented, recreating a user constructed, cross-domain workflow. This will include the upload process for datasets and models, as well as the construction of a workflow with an accompanying parameter set. We will finish with a discussion on the sharing and reproducibility of workflows on the platform.