Infrastructure
The Data Science +X Research Center operates its own cluster infrastructure for scientific and non-commercial projects. It was acquired from own resources and ERDF funds (European Regional Development Fund). The operation takes place in the university computing center.
Software
- Kubernetes cluster with GPU support
- Own Docker registry
- Accessibility for internal and external partners of Forschungszentrum Jülich
- Flexible use of different software stacks, depending on the research task: Spark, Impala, Hive, HDFS, TensorFlow, ...
Hardware
- 2 redundant master nodes
- 15 cluster nodes, 4 x each AMD 6378 (16 cores), 512 GB RAM, local SSD
- Separate main memory database, 4x Intel XEON E7 4890 v2 (15 cores), 1 TB RAM, 10 TB local memory
- Separate GPU node (4x K80)
- Separate GPU node (4x P100)
- Integrated 10GBit - connection of all components
- Approximately 400 TB in the distributed file system (GlusterFS, RAID10 operation)