Computing for extreme conditions
Posted: 27 Jun 2019 | 10:06
The DARE project is addressing the challenges of combining extreme data, extreme computation and extreme complexity in scientific research.
Virtually every scientific domain is experiencing an increase in the volume of data it produces, with growing computational power enabling more complex simulations. Although comparing these simulations with observation can improve models and understanding, it is highly data-intensive.
There is a need for tools that make it easier to reliably formalise data-driven methods that exploit high-end architectures efficiently. DARE is working to empower domain experts to invent and improve methods and models by providing a new platform and working environment. The project’s innovative technology will translate the scientists’ methods into concrete applications that can be deployed and executed on cloud resources offered by European e-infrastructures, as well as in-house institutional platforms and commercial providers.
DARE is initially focusing on seismology and climate research. Working with seismologists at The National Institute of Geophysics and Volcanology (Italy), EPCC has developed interfaces to support the Rapid Ground Motion Assessment (RA) application, which was chosen as our first seismology use case for both its data and computational challenges.
RA aims to model the strong ground motion after large earthquakes to allow rapid assessment of the earthquake’s impact and to support an emergency response. The application requires rapid data analyses, handling multiple data formats, multiple data sources, and the availability of computing and storage resources on demand. Our main objective was to build an application that it is portable and reproducible with minimum need for changes if it runs on different computing resources. To do this we exploited different technologies such as scientific workflows, stream-based data-flow systems, containers, infrastructure orchestrations, notebooks, and cloud platforms.
DARE has created a fluent path from prototyping to production with this new application. It can be run on different computing resources and will scale up automatically without user involvement.
In future, the DARE platform will act as an intermediary between users’ applications (such as RA) and the underlying computing resources. Users will submit their applications through an interface that will automatically deploy the necessary environment on demand to run and monitor the applications, in addition to collecting their provenance and results.
DARE website: project-dare.eu
Rosa Filgueira, EPCC