- 149 participants from 21 countries and four continents
- 17 plenary talks and 21 workshops
- 450 virtual machines, plus 800 physical CPU cores
- 1.5 TB total memory and 10 TB storage
- 25,000 GPU cores
The international GridKa School at the Karlsruhe Institute of Technology (KIT) in Germany celebrated its 13th birthday on 7-11 September with 149 participants from four continents. In 17 plenary talks and 21 parallel tutorials, GridKa School students took a dive into recent trends in big data, virtualization, and modern programming.
Virtualization has become a major theme, especially as computing resources grow increasingly abstract and software-defined. This year's GridKa School focused attention on lightweight virtualization of programs with containers, as well as software-defined networks (SDN) and distributed file systems and databases. Through a talk and tutorial on software-defined data centers, students also got an overview on how to consistently integrate all these resources.
- Programming tutorials took on efficient utilization and parallelization of both CPUs and GPUs.
- The big data section covered topics such as when to use relational and non-relational databases, and how to analyze and process data with various tools.
- The challenges of migrating from IPv4 to IPv6 featured prominently in overview talks by David Kelsey of the Science and Technology Facilities Council (STFC).
- New developments within data-intensive sciences took center stage in a presentation by Peter Wittenburg of the Research Data Alliance (RDA) and Damien LeCarpentier (EUDAT).
- The value of large-scale data processing to simulations and forecasts in fields beyond physics became clear in a talk by the German meteorological service (Deutscher Wetterdienst DWD).
To supply sufficient resources to all participants in the 21 workshops, Preslav Konstantinov and Samuel Ambroj-Perez from the GridKa team set up a cloud-based school on OpenStack. Up to 450 virtual machines (VMs) ran in parallel on 100 hosts to cope with tutorials that demanded up to five VMs per participant.
In addition, KIT's Institute for Data Processing and Electronics (IPE) provided GridKa School with a GPU cluster of about 25,000 GPU cores, and the German Electron Synchrotron (DESY) added further VMs for workshops like a tutorial on site management with Puppet.
While it started as a school on grid computing for high-energy physicists, over the years GridKa School has broadened its spectrum of topics. This year, participants came from scientific backgrounds such as computer science, physics, meteorology, and the medical/biological sciences, with professional experience ranging from graduate students to site administrators.
Despite these differences, participant responses to this year's mix of talks and courses were very positive. The traditional GridKa School tarte flambée social event was also very well received, because after all, school is not only about learning.
Of course, none of this would have been possible without the dedication of the tutors and speakers from various national and international institutions and the members of the GridKa School team from KIT — who have already started to prepare for the 14th GridKa School in 2016. See you next year!