EISCAT-3D: a Data Centric Design for the Extreme Scale Computing

Abstract

The EISCAT-3D experiment will start production in 2018, taking 2PB per year initially up to 3.5 PB per year in its full operations. Data produced at great speeds and volumes at the online will be dumped to the offline, which will produce further user data. This presentation describes the main challenges of the computing model and an overall architecture based in the current prototype of the EISCAT-3D Competence Centre of the EGI Engage project, which has recently adopted the dirac.egi.eu service for Big Data challenges. The prototype is using the DIRAC engine and previous EISCAT data to do fast design of user needs, with a strong data-centric perspective. The main user application is the DIRAC file catalogue, metadata and provenance, which is customized and connected to other DIRAC systems, to cover specific needs of the community, targeting the production EISCAT 3D portal in 2018.