Program - Dag 2

9:00 - SPOR 2

Slurm Update

Køsystemet Slurm er meget udbredt til resource-styring på supercomputere. En opsummering fra det årlige Slurm User Group møde i Madrid vil blive rapporteret, herunder om de nye muligheder i Slurm version 18.08.

13:30 - SPOR 1

Building the future (Nordic) network together

Interesting things are happening in how the NREN community will build their next generation networks. Not only do we take more control of the photonic layer, taking advantage of the disaggregation of the optical network components, we also look into reusing and sharing infrastructures across borders in order to create more capacity, resilience and cost effectiveness in the spirit of doing even more inter-NREN collaboration.

13:30 - SPOR 2

Kulturarvscluster: Reloaded

What is going on at the cultural heritage cluster, what have we learnt in the last twelve months, and where do we want to be in another twelve months?

13:30 - SPOR 3

ERDA: Scalable Storage and Data Management for Science

Scientific data production grows much faster than data from most other areas, and with it the need for affordable scalable solutions to store and analyse the data. SCIENCE at University of Copenhagen has decided to build a scientific storage and data management system in-house; both hardware and software stacks. The resulting system is known as ERDA, and the talk will motivate the features that it provides and argue why we have chosen to build and host a custom system, rather than purchase or rent a generic storage solution.

14:30 - AFSLUTNING

Analysis of big longitudinal data from the healthcare sector

Improvement in more precise treatment of individual patients will in part be driven by the availability of novel molecular data (e.g. genomes, cancer genomes, proteomes), but also by new analysis approaches extracting deep phenotyping signals from existing data. The latter approach works typically from highly heterogeneous data sources: patient record text, laboratory values, images, diagnoses and procedures. An important aspect will be better methods for analysis of longitudinal data in a life-long perspective, leading to better patient stratification and possibly redefinition of what a specific disease actually is. The current disease terminologies are largely defined from symptoms and observations, and not from mechanistic, molecular level insight. Analysis of big longitudinal data from the healthcare sector make entirely change that.

9:00 - SPOR 2

Ceph at DTU Risø

At the beginning of 2018, AIT at DTU Risø decided to set up a ceph cluster as a low-cost large-volume storage system. The cluster is stretched over two locations and its initial purpose is to provide block storage to KVMs and a home file system for our HPC clusters. We will present details about the hardware, deployment (with Open-HPC and docker) and storage configuration of the cluster as well as results of benchmarks. We will finish with a Q/A session and our experience to date in general.

13:30 - SPOR 2

TopWing PRACE project published in Nature

The optimal topology of large structural systems has until now been concerned with the design of individual parts and not that of complete assemblies. However, due to recent advances in numerical algorithms tailored for large scale structural optimization this limitation has now be circumvented. Through a PRACE grant, TopWing, it was possible to test and validate the numerical setup on a real-world design problem, i.e. the design of the supporting structure of an entire wing from a Boeing 777 type aircraft. Besides presenting the optimized wing structure and the HPC resources used for its computations, this talk will focus on the authors general experience with HPC systems, both national and European, including our hopes and wishes for the near future."

11:00 - SPOR 3

A national strategy to support the technical implementation of data management

More and more data services emerge to help scientists to manage their research data. However, most of them are limited to support only one or two stages of the data life cycle, e.g. data archiving and publishing while scientists need tools to manage and track their data from data creation, over data analysis to final publication. Especially the tracking and thus creating a full audit trail of data becomes very difficult if data is spread over different services and storage infrastructures.

Here we present a setup of a data management platform supporting data management throughout the whole data life cycle. The platform integrates existing solutions and provides means to define policies according to which data transits from one stage of the data life cycle and its appropriate services to the next. In that way the platform keeps track of all operations on the data.

Next to providing the technical tools for data management, the Dutch research community is organising support for building and maintaining such solutions. As there is no one size fits all solution to various data management challenges we are working with experts from different research communities to understand specific issues and then provide a customisable and sustainable solution. We will briefly outline how the Netherlands aims to organise RDM experts and harmonise technical developments across institutes.

11:45 - SPOR 2

Scientific output from national supercomputers

High Performance Computing (HPC) has been accessible on a National level for the universities since 2014. This talk gives an overview of the first 290 publications that include National HPC in terms of disciplines, international ranking of journals, collaboration patterns and software tools used for data processing. See more at: https://vidensportal.deic.dk/en/publications

9:00 - SPOR 2

Ansible and OpenHPC at DTU Risø

At DTU research IT Risø the HPC group uses Ansible and OpenHPC for simple and standard deployment of HPC installations and clustered applications. I will introduce the two projects and present details of our related experience in deploying a new HPC cluster.

9:00 - SPOR 3

Multigenerationsregistret - at danne nye registre fra gamle oplysninger

Digitalisering af gamle kirkebøger og matchalgoritmer kan danne et nyt register over familierelationer før det er registreret i CPR-registret.

11:00 - SPOR 1

Forskningsnet for alle

Nettet udbredes til flere institutioner og flere anvendelser.

Tjenesterne bliver også flere og relevante for større målgrupper. Hør om der er uopdagede muligheder for dig og din institution.

11:45 - SPOR 4

Operationelle beredskabsplaner der virker i praksis

Kriseledelse og beredskab er blevet vigtigere i erkendelse af, at vi ikke kan beskytte os mod alle fejl og angreb. Beredskabsplaner er dog ofte for teoretiske og uklare. Her får du konkrete råd og anbefalinger der kan gøre din egen beredskabsplan klasser bedre. Erfaringer fra forsvaret, luftfart, redningsberedskab og talrige virksomheder samles i enkle og klare anvisninger.

9:00 - SPOR 1

Det Kgl. Biblioteks brug af Zoom videokonference og Zoom Rooms

Frank Lund vil med udgangspunkt i fusionen mellem Statsbiblioteket og Det Kongelige Bibliotek fortælle om, hvilken rolle Zoom har spillet i opbygningen af den nye organisation. Det bliver krydret med lidt tekniksnak. Undervejs vil der være en kort demo, hvor deltagerne får mulighed for at forbinde til et Zoom Room på Det Kgl. Bibliotek.

9:00 - SPOR 4

Integritet eller konfidentialitet – to historier fra den virkelige verden med forskellige motiver

Hvor bevæger din data sig hen, når den forlader dine lokaler? Hvem kopierer og ændrer din data - og hvordan? Hvad kan du gøre for at sikre dig mod denne usynlige trussel?

Med udgangspunkt i to meget forskellige historier gennemgås angrebspunkter på fiber og VPN linjer. Der afsluttes med oversigt over forskellige metoder til detektering og sikring af dataforbindelser.

11:00 - SPOR 3

Distributed analysis of meteorological data for wind energy

Both modeled and measured meteorological data is growing in size, which requires researchers to find new approaches to working with that data. In the New European Wind Atlas (NEWA) project, we will have approximately 150 TB of meteorological data that we will want to provide to users in different forms, from timeseries at an aggregated level, to maps of the 30 year mean wind. To provide this data in a timely manner, we will be using the DASK python package. This package works with the rest of the numeric python stack to provide advanced parallelism to scale with our data.

In this presentation, I will provide an overview of the DASK package, describing how it works on some small examples. Then I will provide a couple of examples of how we have used the DASK package to work with the existing subset of NEWA data.

11:00 - SPOR 4

Trusselsbilledet fra Center for Cybersikkerhed (CFCS)

Trusselsvurderingsenhedens vurdering af cybertruslen mod Danmark samt tendenser i trusselsbilledet. Yderligere vil Thorsten kort beskrive Center for Cybersikkerhed og Trusselsvurderingsenheden.

9:00 - SPOR 3

Etablering af et erhvervsstatistisk datavarehus

Forskere har igennem årtier haft mulighed for en internationalt set unik adgang til anonymiserede statistikdata om enkeltpersoner via Danmarks Statistiks forskerserviceordning. Denne mulighed har ikke eksisteret for virksomhedsdata. Med projektet Danish Research Data for the Social Sciences (DRDS), som har fået tilskud fra Uddannelses- og Forskningsministeriets pulje for forskningsinfrastruktur, etableres et erhvervsstatistisk datavarehus, der skal sikre forskere adgang til konsistente mikrodata på tværs af de enkelte erhvervsstatistikker, ligeledes via Danmarks Statistiks forskerserviceordning. Præsentation vil informere om projektets indhold og tidsplan.