BSC Leads Multi-Partner Consortium in Innovative Data Mining Project Across Compute Continuum

22 March 2023

The Barcelona Supercomputing Center (BSC) is the coordinator of the EU-funded EXTRACT project which began on 1 January 2023 bringing together a 10-partner consortium from Spain, France, Italy, Finland, Israel and Switzerland. This three-year project will work to provide a distributed data-mining software platform for extreme data across the compute continuum. It pursues an innovative and holistic approach to data mining workflows across edge, cloud and high-performance computing (HPC) environments and will be validated through two use cases that require extreme data: crisis management in the City of Venice and an astrophysics use case.

Data has become one of the most valuable assets worldwide due to its ubiquity in the thriving technologies of Cyber-Physical Systems (CPS), Internet of Things (IoT) and Artificial Intelligence (AI). While these technologies provide vast data for a variety of applications, deriving value from this raw data requires the ability to extract relevant and secure knowledge that can be used to form advanced decision-making strategies.

The BSC researchers will play a critical role in the project by developing data-driven deployment and scheduling methods required to select the most appropriate computing resource. They will also develop a distributed monitoring architecture capable of securely observing the performance, security, and energy consumption of data-mining workflow execution. Moreover, the BSC will explore various strategies, including AI-based orchestration for deploying and scheduling workflows, to ensure that the various goals are optimized holistically while respecting the constraints imposed by extreme data characteristics.

Current practices and technologies are only able to cope with some data characteristics independently and uniformly. EXTRACT will create a complete edge-cloud-HPC continuum by integrating multiple computing technologies into a unified secure compute-continuum. It will do so by considering the entire data lifecycle, including the collection of data across sources, the mining of accurate and useful knowledge and its consumption.

The EXTRACT platform will be validated in two real-world use-cases, each having distinct extreme data and computing requirements.

BSC researchers will develop the data-driven deployment and scheduling methods needed to select the most appropriate computing resource. This task will address the hereto unaddressed challenge of orchestration on the edge-cloud-HPC continuum. It will include ensuring that orchestration technologies are explicitly aware of extreme data characteristics and workflow description.

The BSC will also develop a distributed monitoring architecture that will be capable of securely observing the performance, security, and energy consumption of data-mining workflow execution. To ensure that various goals are optimised holistically while respecting constraints imposed by extreme data characteristics, BSC will explore various strategies including AI-based orchestration for deploying and scheduling workflows.

Eduardo Quiñones, established researcher at the Barcelona Supercomputing Center and EXTRACT coordinator, is confident that: “By seamlessly integrating major open-source AI and Big Data frameworks, EXTRACT technology will contribute to providing the technological solutions Europe needs to effectively deal with extreme data. It will go beyond facilitating the wider and more effective use of data to reinforce Europe's ability to manage urgent societal challenges.”

About EXTRACT

The EXTRACT project (A distributed data-mining software platform for extreme data across the compute continuum) is funded under Horizon Research and Innovation Action number 101093110. The project began on 1 January 2023 and will end 31 December 2025. The consortium, formed of 10 partners, is coordinated by the Barcelona Supercomputing Center (BSC). Consortium members include: Ikerlan (Spain), Universitat Rovira I Virgili (Spain), Observatoire de Paris (France), the Centre National de la Recherce Scientifique (France), Université Paris Cité (France), Logos Ricerca e Innovazione (Italy), City of Venice (Italy), Binare (Finland), Mathema srl (Italy), IBM Israel (Israel), sixsq (Switzerland).

 

  • Caption: EXTRACT consortium meets in Barcelona for its kick-off meeting hosted at the home of the Barcelona Supercomputing Center´s famous MareNostrum supercomputer.