After being cancelled last year due to COVID-19, the annual DeiC Conference returned again this year. Three of our employees give a report of the event and offer their thoughts on the variety of talks.
Dr. Martin Rasmus Lundquist Hansen, Team Leader Research Infrastructure
“For me the DeiC conference was an opportunity to hear about current topics relevant for the continued development of IT infrastructure at the Danish universities. While I attended a variety of talks, I found the talks about security and the continued expansion and development of the Danish Research Network the most interesting.
In recent years security has become an integral part of any development cycle, and for good reason. The need for continuously enhanced security measures is needed to keep attackers out, which was also emphasized by some of the talks, where security specalists explained and highlighted some of the day-to-day challanges and tasks that are needed to keep the network secure, not only locally, but also on the internet in general. Since everything is connected, keeping the internet secure also helps keeping the local users secure.
During the covid lockdown, the Danish Research Network started to play an even more pivotal role in the daily work of all researchers and students who suddenly were forced to work remotely. In fact, during the last two years the network has been expanded to include more than 100 new locations, such that more and more institutions have a fast and reliable connection to all other institutions, both nationally, but also in the rest of Europe and Scandinavia. As part of the development, the network is also in the process of being upgraded to 100 Gbps between the universities and other core institutions, with 400 Gbps connections being planned within the next 3-4 years. These upgrades are important and needed to efficiently move the increasing amount of research data being produced at the national HPC facilities.
Last but not least, the conference was also a great opportunity to talk with colleagues from other universities, some of which I had never met in person because of the recent lockdowns. I also had interesting discussions with some of the vendors present at the conference, talking about new products and solutions that might be of interest to us in the future.”
Brian Ravn, Software Architect
“After more than a year without too many social gatherings, the two days at DeiC conference was a nice change. The conference was an opportunity to hear about the current state and future plans for HPC and eScience in Denmark and in Europe.
Meeting colleagues from other parts of Denmark was great, and it was interesting to hear about their work, and get some feedback on our work on UCloud.
As a developer I was mostly drawn to the talks on HPC, eScience and data ethics. I found the success-stories, i.e. how HPC systems had been used for various projects in practice, were some of the most interesting.
Being attentive of what researchers and other users of HPC find valuable and what problems they face, is extremely important if we want to keep improving UCloud in the future.”
Dr. Emiliano Molinaro, Team Leader Research Support
“This year’s edition of the DeiC conference featured several talks focused on data management and big data processing, which were of particular interest for me and generated new ideas.
In particular, there was a long discussion, divided into several panel sessions, on different aspects of FAIR data principles: from their historical development to the implementation of a dedicated national strategy, as well as their applications in the private sector. The FAIR acronym stands for data which is findable, accessible, interoperable, and reusable. These high-level principles were systematically defined for the first time in 2016.
There are two big challenges for the implementation of any of these principles: the first is to create machine-actionable metadata, which is based on common minimum specifications, across different domains. The second is to find a way to scale the production of metadata.
While there is not yet a common practice on how to solve these two challenges so far, some use cases were discussed by some of the speakers at the conference. On the one hand, the challenge of FAIR data at scale is to provide a unified access for researchers located in multiple distinct locations to a large number of heterogeneous datasets, often created by automated pipelines. On the other hand, as in the case of health sciences, the sensitive nature of patient data makes it difficult to make data openly available. In such cases, one strategy discussed was to implement a federated data model, where data are kept locally and only aggregated data are collected, while both data and metadata use identifiers and vocabularies that follow clear FAIR principles.
In addition, an interesting debate was also held on the reproducibility of the research results, in relation to FAIR data and the problem of data versioning.
These and other ideas have given me new inspiration about how to use our cloud services to help researchers at SDU, but also nationwide, automatically implement a FAIR-oriented data management plan.”