At SURF, we aim to learn from other institutions and HPC centres about the technologies they have adopted, their use cases, and envisioned strategic impacts. Some of the previous expeditions included visits to CSC Finland & NextCloud in Berlin. In this iteration, together with, Sagar Dolas, Raymond Oonk, Monique Denissen, and SURF colleagues, we organised a two-day visit to CSCS, the Swiss National Supercomputing Centre. These two days were filled with deeply engaged discussions, dialogue, shared perspectives, and knowledge exchange. It was very inspiring to meet the CSCS colleagues and share our mutual ambition to support and serve science with large-scale computing.
The most critical takeaway for our roadmap at SURF is the transition from a hardware-centric model to a "Service-to-science" approach. Seeing a concrete vision of CSCS’s architecture, which balances flexibility and performance, has already proven valuable in sharpening our ongoing conversations and aligning our current thinking within SURF.
For our visit, we were represented by different SURF teams, e.g., Cloud, Data Management, HPC, Data Processing, and Future Technologies. Because of this diversity, we delved deeply into specific and technical discussions. For example, we learned how CSCS facilitates computing environments for specific research communities, such as climate modelling and AI, by abstracting the Alps infrastructure into Versatile Clusters. The different innovation initiatives and programmes are currently exploring this path, using Alpernetes as a blueprint for our own orchestration.
We also had time to learn more about CSCS's strategic goals and how they align with or differ from SURF's. As a national supercomputing centre, its scope is somewhat smaller than SURF’s. However, they face many of the same challenges in enabling workflows. CSCS addresses this by focusing on specific but common challenges across research communities. This is also reflected in their organisational structure, which emphasises agile capabilities and flexibility. They follow the logic of value streams and organise their teams around thematic capabilities rather than strict hierarchical management lines known as working structure. Each member can be part of multiple working structures, which makes knowledge exchange frictionless and organic. A working structure is end-to-end responsible for its components. There are no operations; you either support other Working Structures or a community. Also, they work less in projects but are more focused on key topics and communities. This arrangement makes them unique, valuable, and at the forefront of innovation in computing, enabling & accelerating research.
On the last day, we also visited the supercomputer and learned more about the physical integration of the data centre. The centre relies on Lake Lugano for cooling and then feeds that thermal energy back into the city's heating system. This sets a high standard for environmental circularity that aligns perfectly with SURF’s sustainability goals. In addition to two days of work, we had the opportunity to connect socially in beautiful Lugano. On day one, we met for a shared dinner with the CSCS team, where the day's discussions continued in a more relaxed setting.
We firmly believe that we will continue this collaboration, mutually strengthening each other and reaching our collective goals and ambitions.
A special thanks to Maxime Martinasso, Pim Witlox, and Pablo Fernandez for hosting us and giving us a warm welcome.
From SURF, we thank following colleagues who participated and contributed to the sessions with presentation, discussions and ideas: Arijit Laik, Jorik van Kemenade, Kristen Lutz, Maithili Kalamkar, Manuel Torres Rodrigues, Tim Kok, Coen Schrijvers, Guiseppe Gianquitto, Ron Trompert, Robert Jan Schlimbach, Martijn Kruiten, Nikolaos Parasyris.
0 Praat mee