First stress test of the PhenoMeNal application stack running on the de.NBI cloud

Home  /  Tools  /  First stress test of the PhenoMeNal application stack running on the de.NBI cloud

The CloudMet2017 summer school organised by CRS4 in Sardinia, brought  together participants and trainers from eight different countries. The event proved to be an opportunity for participants to learn a lot about metabolomics data analysis in the cloud, and the PhenoMeNal consoritum  was also glad to see that our infrastructure(s) were chugging along while 30+ participants followed the trainers and ran the hands-on tutorials in the cloud.

In addition to the Public PhenoMeNal instance hosted at EMBL-EBI and ad hoc private instances on resources donated by the AWS Cloud Credits for Research program, the participants ran multivariate statistics in Jupyter Notebooks that were deployed in the de.NBI Cloud.

Resources used on the de.NBI cloud

The virtual network setup in OpenStack

Thanks to the automated cloud deployment, scaling from just one Jupyter instance up to 40 – one for each participant and trainer – was a matter executing just a single command line. Using the graphical monitoring features configured on our de.NBI deployment we were able to follow the load as the participants went through the individual analysis steps:

Kristian Peters explained RDA to 30 participants at 16:54. And Wilks statistics at 17:03.

Session over!

After the session, participants learned how to download their own analysis as a Jupyter R notebook, or as HTML or PDF documents if they preferred. If this had been their own experiment, they could add these reports as-is to the supplemental material of their next paper.

In summary, we were happy to see the PhenoMeNal infrastructure pass its first stress tests and contribute to make CloudMET 2017 an excellent experience for participants and trainers alike.

Author: Dr. Steffen Neumann

Comments are closed.