12.04.2021 | IT-Infrastructure
Accelerating Data Science With New Computing Resources
With its new IT infrastructure, the DSC offers flexible computing capacities to perform complex algorithms and data analyses in research and teaching.
Data science, particularly AI-based analysis methods, require powerful computational capacities beyond the means of a classical office computer. Accordingly, the Data Science Center (DSC) at the University of Bremen has established a new IT-infrastructure enabling fast and effective data analyses as well as the implementation of machine learning experiments. This is based on the three pillars of High-Performance Computing (HPC)
, and Virtualization
In the area of HPC
, the DSC is equipped with powerful GPU servers that use three NVIDIA V100 GPUs to boost the speed and performance of data analytics. In fact, one NVIDIA V100 GPU provides the performance of up to 32 CPUs, making it one of the most advanced data center GPUs ever built to accelerate data science and AI. In addition, the DSC supports multiple software environments, including the most popular languages such as Python or R and their commonly used libraries and packages, to meet the diverse needs of our users.
Data science involves large amounts of data and to effectively process such big data, a large bandwidth is required. Accordingly, the servers of the DSC are interconnected through state-of-the-art InfiniBand technology permitting high-speed data transmission with rates of up to 100 gigabits per second. As part of the pillar Storage
, the DSC is currently equipped with 5 SSDs providing a total storage capacity of 5.75 TB for operational use, which can be scaled up to 24 SSDs depending on the needs of the users.
As technology gets more complex, its security becomes more elaborate and crucial. For maximum user friendliness, we will offer a same sign-on authentication for user accounts of the University of Bremen. In the area of Virtualization
, the DSC is equipped with a Virtual Machine (VM) server, allowing us to offer customized resources depending on the user’s needs. Our VM service can for instance be used to implement Web crawlers or applications like R Studio Server.
In summary, the DSC offers optimal and flexible resources for running computationally intensive algorithms and data analysis in data-intensive research and teaching. If you are interested in using our computational resources, please send us an e-mail to discuss the possibilities.
Are you interested in using our IT-infrastructure?
Dr. Lena Steinmann
+49 (421) 218 - 63941