Innovative IT infrastructure for research and science
Good prospects for 2025: The LRZ brings new resources into everyday life and works on its IT services. Photo: Foyez Ullah/Adobe
BayernKI, HammerHAI, Blue Lion – LRZ had announced quite a few systems in 2024. These will be implemented at the LRZ in 2025. More on what researchers can look forward to in 2025 at the LRZ.
It has already found its place on the 2nd floor of the computer cube: CoolMUC-4 contains 193 Xeon Platinum processors from Intel, works with the Suse Linux Enterprise Server 15 SP16 operating system and will be expanded with Graphics Processing Units (GPU) in the course of the year. Researchers will use the new Linux cluster at the Leibniz Supercomputing Centre (LRZ) to perform classic modelling and simulation tasks and – after the GPU expansion– combine these with artificial intelligence (AI) methods. This will allow models to be simulated faster than before, for example to explore weather or other natural phenomena more thoroughly.
CoolMUC-4 is not the only system that will go into operation at the LRZ this year. If 2024 will go down in the annals of the LRZ as a year of announcements of new systems and resources and centrally organised IT infrastructures, 2025 will see the implementation of all these ambitious projects on the agenda: innovative technologies have been procured for scientific computing – in some cases with partner organisations – and further quantum systems will be put into everyday operation. In addition, the LRZ is optimising many of its IT services and continuing to improving its resilience in order to resolve failures of its services and computing resources more quickly and efficiently and in order to better ward of attacks on its infrstructure. "Scientists can expect us to ensure that our IT services continue to function reliably and that their information is protected,” Prof Helmut Reiser, the Deputy Director of the LRZ, says summarising the LRZ's most important goal for the year: The second re-certification of the data centre in the disciplines of IT service management (ISO/IEC 20.000) and information security (ISO/IEC 27.001) is scheduled for summer 2025. And LRZ employees are already busy revising the documentation of internal processes so that external auditors can assess them and - hopefully - issue the coveted certificates.
Supercomputing, artificial intelligence and quantum systems
As part of its Hightech Agenda, the Freestate of Bavaria has created more than 130 AI professorships and is now also investing in an AI infrastructure called BayernKI, which is gradually being established and expanded. The high-performance systems required for this have been set up at the LRZ and the Erlangen High Performance Computing Centre NHR@FAU. In combination with consulting, support and courses for AI experts, BayernKI already offers fast and flexibly scalable access for researchers and students from all disciplines who are researching and developing AI models.
As in Bavaria, the aim is to create comprehensible, useful, energy-efficient and data protection-compliant AI and language models for all areas of society in Europe. To this end, EuroHPC Joint Undertaking is funding seven European AI factories, one of which is in Germany: the HammerHAI project, which is co-financed by several German ministries, will soon offer scientists, companies, start-ups and the public sector an innovative supercomputing infrastructure optimised for AI methods. The systems are installed at the High Performance Computing Centre Stuttgart (HLRS). The LRZ will support the endeavour with consulting and user assistance leveraging synergies with BayernKI. For example, it supports researchers in selecting suitable AI models, integrating them into the systems and providing workshops and training.
In addition to AI, hopes are pinned on quantum computing and quantum processors as supercomputer accelerators: In summer 2024, the LRZ became the first data centre in the world to successfully integrate the Q-Exa system based on superconducting circuits with 20 qubits into a supercomputer. The Euro-Q-Exa project builds on this ground-breaking achievement: To ensure that Europe's research community can also quickly reckon with this future technology, the LRZ is integrating a more powerful quantum processor into its supercomputers on behalf of EuroHPC Joint Undertaking. In an initial phase, it will work with 50 qubits - in this category, the quantum advantage can become visible, which not only accelerates calculations and simulations, but also adds further new calculation and analysis methods to the scientific tool set. Researchers can use Q-Exa for their projects from spring, Euro-Q-Exa will starting in autumn 2025.
New systems for everyday research
At the same time, scientists can look forward to another quantum technology: by the summer, the LRZ will open a computer in which an ion trap and laser light make 20 electrically charged atoms usable as qubits for computing. The system from Alpine Quantum Technologies (AQT) has been set up faster than planned, and LRZ specialists are currently working intensively on it so that users can soon be actively supported in their work. Which calculations does this system carry out with particular accuracy? And how much effort does it take to calculate with it? The first reports on how quantum technologies differ in daily use and where users need special support can be expected in 2025: We are very much looking forward to these results.
Additionally, the anticipation is growing for the next supercomputer and successor to SuperMUC-NG: the contracts to build Blue Lion were signed with Hewlett Packard Enterprise (HPE) at the end of 2024. Although the exascale-ready system will not be installed until 2026, practical preparations are already starting this year: Small lions will be stalking the LRZ until the summer. On ‘Blue Cub’, researchers can port codes and customise them with the help of LRZ experts and HPE Labs Europe. Suitable research projects will be identified in the spring. The mentors from Computational X Support (CXS) and the future administrators will familiarise themselves with the new technology on a second migration system. Tangible research results will take centre stage at the SuperMUC-NG User Workshop in spring 2025: researchers will show science and results that SuperMUC-NG has enabled and discuss the insights they have gained in the process.
Secure and efficient services
The (further) development of the LRZ's IT services can be summarised by the keywords resilience and energy efficiency. This strategy provides users and master users who are responsible for IT provision at university chairs and institutes with greater security and convenience: CoolMUC-4, for example, will receive a new ID management system over the course of the year, which is no longer linked to individual projects but to users and is easier to access. LRZ experts are preparing more training videos to demonstrate the benefits of group and cross-university collaboration of the BayernCollab documentation platform, which is rolled out even further. With the introduction of Veeam Cloud Connect from around summer 2025, institutions will be able to create an additional, automated backup at the LRZ that is protected against malware and for which data is encrypted. This protects user data in the event that the universities' backup systems fail.
The LRZ offers BayernShare for the exchange of files; the service will celebrate its 10th anniversary in 2025 and will be additionally protected: Another "disaster system" is currently being installed at the Erlangen Regional Computer Centre (RRZE), which will store all data once again as a precaution. Should servers at the LRZ in Garching fail, the data can be made available again at lightning speed via a fast connection to Erlangen's RRZE. By autumn 2025, a second X-Win link to the German National Research and Education Network will also be set up, which is deliberately not located on the research campus for resiliency reasons and will take over data traffic from the Münchner Forschungsnetz (MWN) should its back-up fail. This makes (wireless) communication even more secure and reliable.
When installing and operating IT infrastructure, energy efficiency and the resource-saving operation of data centres is becoming increasingly important, also for cost reasons: in 2025, the in-house photovoltaic system will supply energy for the LRZ offices for the first time. Like SuperMUC-NG and CoolMUC-4, Blue Lion is also cooled with warm water only and no needs for fans. And the LRZ is already preparing for the next certificate: The European environmental management system EMAS documents an organisation's consumption of resources and is intended to help all areas use materials, equipment and energy more sparingly. A first report will be produced by the end of 2025 with more to come in 2026. (vs)