Storage infrastructures to support advanced scientific workflows. Towards research data management aware storage infrastructures

DSpace Repository


Dokumentart: ConferencePaper
Date: 2019-04
Language: English
Faculty: 7 Mathematisch-Naturwissenschaftliche Fakultät
Department: Informatik
DDC Classifikation: 004 - Data processing and computer science
Keywords: Hochleistungsrechnen
Other Keywords:
Research Data Management
Large Scale Research Infrastructures
Storage Governance
Long Term Data Publication
Show full item record


The operators of the federated research infrastructures at the involved HPC computer centers face the challenge of how to provide storage services in an increasingly diverse landscape. Large data sets are often created on one system and computed or visualized on a different one. Therefore cooperation across institutional boundaries becomes a significant factor in modern research. Traditional HPC workflows assume certain preliminaries like POSIX file systems which cannot be changed on a whim. A modern research data management aware storage system needs to bridge from the existing landscape of network file systems into a world of flexible scientific workflows and data management. In addition to the integration of large scale object storage concepts, the long term identification of data sets, their owners, and the definition of necessary meta data becomes a challenge. No existing storage solution on the market meets all of the requirements, and thus the bwHPC-S5 project must implement these features. The joint procurement and later operation of the system will deepen the cooperation between the involved computer centers and communities. The transition to this new system will need to be organized together with the scientific communities being shareholders in the storage system. Finally, the created storage infrastructures have to fit well into the growing Research Data Repositories landscape.

This item appears in the following Collection(s)