Upgrade to Pro — share decks privately, control downloads, hide ads and more …

Reperforming a Nobel Prize Discovery on Kubernetes

Reperforming a Nobel Prize Discovery on Kubernetes

Efee9430e5ca7dec19b1436c5782b477?s=128

Ricardo Rocha

May 21, 2019
Tweet

Transcript

  1. None
  2. 2

  3. None
  4. 4

  5. None
  6. None
  7. }

  8. None
  9. None
  10. None
  11. None
  12. None
  13. None
  14. The Standard Model: the most successful scientific theory … but

    it had a missing, but crucial puzzle piece: the Higgs boson. It helps explain why particles have mass. When we have a powerful enough collider we can create the Higgs boson (remember E~=mc²) But we cannot see it directly, because it instantly decays, e.g. into 4 particles. H
  15. 15

  16. Demo Time

  17. Sim Higgs Background Background Real Data Event Data Summary Data

    Make Plot! 20k+ Core K8s Clusters 70 TB of Physics Data ~25000 Files
  18. OpenStack Magnum 70 TB Dataset Job Results Interactive Visualization Aggregation

    25000 Kubernetes Jobs
  19. Cluster on GKE Max 25000 Cores Single Region, 3 Zones

    70 TB Dataset Job Results Interactive Visualization Aggregation 25000 Kubernetes Jobs
  20. None
  21. Compute like it’s 2010

  22. Containers give us Reproducibility in Space and Time • re-run

    code in the future in a different data-center • Gamechanger for Open Science
  23. Anyone can (re-)discover the Higgs boson!

  24. Lukas Heinrich @lukasheinrich_ Ricardo Rocha @ahcorporto Clemens Lange @clelange Thomas

    Hartland github: @tghartland THANKS!