UvA HPC course 2019-01-30

UvA logo

SURFsara logo

General Information

SURFsara has been operating the HPC Cloud as an Infrastructure as a Service (IaaS) for several years. The HPC Cloud service offers users access to fast computing nodes and high-performant big storage volumes through a powerful graphical user interface. The underlying cluster Oort is managed by the OpenNebula cloud management software.

This workshop aims to introduce the participants to the HPC Cloud, by providing a set of lectures and hands-on examples to be run on the Oort infrastructure. The session is part of the UvA HPC Course which covers various topics related to HPC and Big Data.

Agenda:

30th January 2019, UvA, Science Park SP G0.23-G0.25

  • 09:00 Welcome
  • 09:05 Presentation and Demo: HPC Cloud at SURFsara (Markus van Dijk)
  • 09:45 Hands-on: Parts A & B (at your discression, also C)
  • 12:00 Lunch break
  • 13:00 Morning recap
  • 13:05 Presentation and Demo: Cloudifying - parallelism & API (Ander Astudillo)
  • 13:45 Hands-on: Part C & Extras
  • 17:00 End

Preparation

Participants are expected to bring a laptop with Internet connection and a browser installed. Prior experience with Linux commands will help but not a prerequisite.

For this workshop you will need:

NOTE for Windows users:

For the Extras you may need to start a graphical window in your laptop that is connected to your HPC Cloud machine. To make this work, you need to adjust the following:

  • Open GitBash (or git for windows).
  • In the terminal that appears, copy and paste the following commands:
# press enter to submit each command
echo "export DISPLAY=localhost:0.0" >> $HOME/.bashrc
source $HOME/.bashrc

Hands-on

The hands-on tutorial consists of three parts:

Presentations

Assignment

Download the assignment

Master students eligible to earn ECTS must follow all the lectures and participate on all workshops associated to UvA HPC Course. Access to the HPC Cloud service will remain available to the participants for 1 week after the course. The HPC Cloud course assignment must be done in pairs, and it needs to be delivered by the end of the 1-week period to Adam Bellowum. Adam Belloum will then collect all assignments and make them available to the HPC Cloud team for evaluation.

Thus, make sure you have submitted your assignment before 23:59h on Wednesday one week from today!