Zitationsvorschlag

von Suchodoletz, Dirk et al.: Lessons learned from Virtualized Research Environments in today’s scientific compute infrastructures, in Heuveline, Vincent, Gebhart, Fabian und Mohammadianbisheh, Nina (Hrsg.): E-Science-Tage 2019: Data to Knowledge, Heidelberg: heiBOOKS, 2020, S. 68–81. https://doi.org/10.11588/heibooks.598.c8418

Identifier (Buch)

ISBN 978-3-948083-14-4 (PDF)
ISBN 978-3-948083-15-1 (Softcover)

Veröffentlicht

26.03.2020

Autor/innen

Dirk von Suchodoletz, Jonathan Bauer, Oleg Zharkov, Susanne Mocken, Björn Grüning

Lessons learned from Virtualized Research Environments in today’s scientific compute infrastructures

The Virtual Open Science Collaboration Environment project (ViCE) aimed to promote Virtualized Research Environments (VRE) to be transparently used on various research infrastructures available in Baden-Württemberg. VREs provide researchers with more freedom and flexibility using infrastructures for research and teaching ranging from high performance computing (HPC) and cloud resources to lecture PC pools. The project managed to shape new future operational models of HPC clusters and scientific clouds and to separate contradictory demands regarding software environments. The project reached varying results ranging from a rather broad uptake in the domain of the simpler virtual teaching and working environments for desktop operation compared to the more complex scientific workflows characterized by further external dependencies. Requirements like special filesystem access, a fast message passing interface or the use of special purpose hardware like graphics processing units limit the flexibility of the VRE approach to certain degrees. VREs formalize the abstraction of (complex) scientific workflows from the underlying hardware to  make them more versatile, exchangeable and both archivable and reusable in the long run. Abstraction helps to complement the research data management of results and primary data sets in the future. The broader application of VREs directly relates to the business and operation models of the large scale research infrastructures in Baden-Württemberg like bwHPC and bwCloud. The gained technical flexibility is not necessarily matched to well-established financing and compensation models for the infrastructure providers.