Skip to content

Technology and concept ABC

The IDAF consists of differnt parts, and there is also some concepts involved. Here is a (non exhaustive) list of technologies and concepts.

  • AFS: the well known global file system is used as home directory for the NAF.
  • batch: at the technological and conceptual heart of the IDAF are batch systems that offer scalable resource provisioning
  • CVMFS: the CERN Virtual Machine FileSystem is used to deploy software for some user communities
  • dCache: while batch systems are at the heart of the IDAF, dCache is the brain of the IDAF: It holds most of the data, serves as data exchange and management platform ... and is actively developped at desy
  • GPFS: another storage and data management platform, used for high performance data access, which includes connection to the online data taking
  • Grid: the IDAF federates into worldwide job and data management workflows via Grid technologies.
  • HPC: the IDAF component Maxwell uses technologies and concepts from the HPC (High Performance Computing) world, e.g. high bandwith, low latency network InfiniBand, whole node scheduling and message passing interface.
  • HTCondor: a batch system, also called a scheduler, optimized for high throughput computing, and used in the NAF and Grid components. Developped by great people in Wisconsin.
  • interactivity: we know that data analysis cannot be scheduled to the last degree. We offer interactivity, both in the access methods, and in resource provisioning.
  • Jupyter: you have a browser? you got access to the full IDAF power! We offer Jupyter notebooks, and have them integrated into the scaling batch system transparently - for your convenience and ease of utilization.
  • Kerberos: An network authentication protocol, used within the IDAF as well as one access method to the IDAF. More and more federated methods are being used for the AAI step - if you are a remote user of the IDAF, you already have an account at your home institution, why should you need another one just for the IDAF?
  • Linux: the ubiquitous operating system basis for all services we run.
  • Monitoring: the IDAF is composed by several subsystems. We perform monitoring of the components at all level, to guarantee stable operation, have an empiric basis for future optimizations and decisions, and eventually troubleshooting.
  • NAF: the National Analysis Facility is an analysis facility for German HEP scientists as well as the global BELLE II community.
  • Open: Whenever possible, we utilize software that is Open Source, or at least Open Standards. When we see that we can contribute back the Open Source community, we do so. We aim to enable Open Data - as well as honoring the F.A.I.R. principles.
  • POSIX: A standard defined four decades ago, which still governs many things we do. Like how storage is offered, and what storage can and cannot do.
  • queues and quotas: the IDAF processes a lot of jobs, and stores hugs amount of data. To steer these jobs, the usage of queues and quotas in inevitable. We try however to make these as flexible as possible, so that they get into your science workflows as little as possible.
  • resources: some numbers from 2024: The IDAF constists of around 60.000 CPU cores and nearly 400 GPUs. The 1500 compute servers offer around 600 TB RAM. The storage systems offer around 150 PByte disk on dCache and over 60 PByter disk on GPFS.
  • SLURM: another batch system or scheduling system, optimized for the usage in the High Performance Computing world.
  • Support: we know that you are doing complicated science, and want to make best possible usage of the IDAF to boost your science. We are there to support you at all levels, such that you can fully focus on your work.
  • TensorFlow is one example of an application and application framework being offered on the IDAF. We provide a comprehensive set of scientific software integrated into the IDAF setup, such that you can directly start with your analysis.
  • Users: our motivation for running the IDAF: offering the best possible service for our users! They come from different communities, have different needs ... we are confident, the IDAF is the place for you!
  • virtual environments: The IDAF supports container technologies for offering users and communities their own, portable execution environments for their analysis.
  • WLCG: we are federated into the Worldwide LHC Computing Grid.
  • XFEL: the European XFEL is the single largest experiment the IDAF supports, and the biggest contributor to its hardware resources.
  • sustainabilitY: we do not forget about sustainabilty, in the end, there is no computing on a dead planet. We try to reduce our CO2 footprint at different levels, e.g. user awareness and training, optimizing computer availability and performance in accordance to green energy availability, or rethinking server and computing center lifecycle.
  • Zookeeper: Apache Zookeeper is a framework to manage distributed configuration files, which the IDAF uses for the dCache storage system. This is just another example on how the IDAF components utilize state-of-the-art open source and industry tools to offer a stable, scaling service for users.