Enabling dynamic and intelligent workflows for HPC, data analytics, and AI convergence

• DOI: 10.1016/j.future.2022.04.014
• Publication Date: 2022-04-27T15:35:54Z
• Authors: Jorge Ejarque; Rosa M. Badia; Loïc Albertin; Giovanni Aloisio; Enrico Baglione; Yolanda Becerra; Stefan Boschert; Julian R. Berlin; Alessandro D’Anca; Donatello Elia; François Exertier; Sandro Fiore; José Flich; Arnau Folch; Steven J. Gibbons; Nikolay Koldunov; Francesc Lordan; Stefano Lorito; Finn Løvholt; Jorge Macías; Fabrizio Marozzo; Alberto Michelini; Marisol Monterrub...; Marta Pienkowska; Josep de la Puente; Anna Queralt; Enrique S. Quinta...; Juan E. Rodríguez; Fabrizio Romano; Riccardo Rossi; Jedrzej Rybicki; Miroslaw Kupczyk; Jacopo Selva; Domenico Talia; Roberto Tonini; Paolo Trunfio; Manuela Volpe
• Keywords: FOS: Computer and information sciences; Workflow development; HPC-DA-AI convergence; Parallel programming; Parallel programming (Computer science); 02 engineering and technology; Programació en paral·lel (Informàtica); 01 natural sciences; Àrees temàtiques de la UPC::Informàtica::Arquitectura de computadors; 0103 physical sciences; Computer Science - Distributed; 0202 electrical engineering, electronic engineering, information engineering; Electronic data processing -- Distributed processing; High performance computing; Distributed computing; Parallel Programming; HPC-DA-AI convergence; Workflow development; Workflow orchestration; :Informàtica::Arquitectura de computadors [Àrees temàtiques de la UPC]; Workflow orchestration; Parallel; Distributed computing; High performance computing; Distributed computing; Parallel programming; HPC-DA-AI convergence; Workflow development; Workflow orchestration; Computer Science - Distributed, Parallel, and Cluster Computing; and Cluster Computing; High performance computing; Distributed, Parallel, and Cluster Computing (cs.DC); Càlcul intensiu (Informàtica); info:eu-repo/classification/ddc/004; Processament distribuït de dades

The evolution of High-Performance Computing (HPC) platforms enables the design and execution of progressively larger and more complex workflow applications in these systems. The complexity comes not only from the number of elements that compose the workflows but also from the type of computations they perform. While traditional HPC workflows target simulations and modelling of physical phenomena, current needs require in addition data analytics (DA) and artificial intelligence (AI) tasks. However, the development of these workflows is hampered by the lack of proper programming models and environments that support the integration of HPC, DA, and AI, as well as the lack of tools to easily deploy and execute the workflows in HPC systems. To progress in this direction, this paper presents use cases where complex workflows are required and investigates the main issues to be addressed for the HPC/DA/AI convergence. Based on this study, the paper identifies the challenges of a new workflow platform to manage complex workflows. Finally, it proposes a development approach for such a workflow platform addressing these challenges in two directions: first, by defining a software stack that provides the functionalities to manage these complex workflows; and second, by proposing the HPC Workflow as a Service (HPCWaaS) paradigm, which leverages the software stack to facilitate the reusability of complex workflows in federated HPC infrastructures. Proposals presented in this work are subject to study and development as part of the EuroHPC eFlows4HPC project.