Skip to main content

Which are the computing resources of the WEkEO JupyterHub?

Come discover the computing resources and working environments available in the WEkEO JupyterHub!

David Bina avatar
Written by David Bina
Updated yesterday

Context


Among all services provided by the WEkEO DIAS, the WEkEO JupyterHub is available for free for all users, you simply need a WEkEO account (sign up!).

We'll see in this article the different computing resources you'll find on the JupyterHub behind each server option.

Computing resources


When accessing the JupyterHub, you can choose between these Server Options:

  • Machine Learning

  • Earth Observation Tools

  • Machine Learning (GPU)

Each Server Option has its own computing resources and tools.

Machine Learning

The Machine Learning server is more data science oriented, mostly used to process data and run more complex computation. This server offers 3.5 CPU and 14 GB of RAM, with the following tools pre-installed:

  • Python: with a set of kernels based on Conda environments:

    • base: the default environment

    • climaf: an environment with the climaf library and more

    • pcmdi: an environment with the pcmdi library and more

    • wekeolab: an environment with all basics and useful packages needed by the majority of users, including the hda Python package

  • R: a set of data science librairies is included. You can open a terminal and list their versions with the following command:

    R -e 'data.frame(installed.packages())[,c("Package", "Version")]'

  • R Studio: available from the JupyterHub home page, clicking on the RStudio button. A new browser window opens in which RStudio can be used, the working directory and files are visible within RStudio

  • Julia: a Julia environment is provided, with a number of libraries installed. You can list the installed packages by running Julia in your terminal with the julia command, then pressing the "]" key and running the status command

💡WEkEO Pro Tip: we recommend the Machine Learning server during the WEkEO trainings and webinars! 😄

Earth Observation Tools

The Earth Observation Tools server is designed with the simple purpose of visualizing data and results. As the Machine Learning server, it offers 3.5 CPU and 14 GB of RAM, with the following tools pre-installed:

  • Python: with a set of kernels based on Conda environments:

    • base: the default environment

    • wekeolab: an environment with all basics and useful packages needed by the majority of users, including the hda Python package

  • R: a set of data science librairies is included. You can open a terminal and list their versions with the following command:

    R -e 'data.frame(installed.packages())[,c("Package", "Version")]'

  • R Studio: available from the JupyterHub home page, clicking on the RStudio button. A new browser window opens in which RStudio can be used, the working directory and files are visible within RStudio

  • Julia: a Julia environment is provided, with a number of libraries installed. You can list the installed packages by running Julia in your terminal with the julia command, then pressing the "]" key and running the status command

⚠️ Important remarks for Machine Learning and Earth Observation Tools

  • Kernel wekeolab: all the necessary packages for downloading and post-processing WEkEO data are already installed in the wekeolab environment, including hda. However, it is also possible to create a personal environment!

  • R package updates: R packages installed in WEkEO JupyterHub are managed centrally and cannot be updated by users. This ensures compatibility and stability within the environment.

🔍 If you require a specific version of an R package or need additional R packages that are not currently available, please don’t hesitate to contact WEkEO User Support.

Machine Learning (GPU)

The Machine Learning (GPU) server on WEkEO JupyterHub provides a shared environment tailored for lightweight machine learning and deep learning workloads.

Each user session runs on a node equipped with a GPU offering 6 GB of memory, which is ideal for prototyping and running moderately demanding models. Resources are allocated with a limit of 3 CPU cores and 12 GB of RAM per user session.

The following tools are installed on the Machine Learning (GPU) server:

  • Python: with a kernel based on Conda environments:

    • base: the default environment

  • R: a set of data science librairies is included. You can open a terminal and list their versions with the following command:

    R -e 'data.frame(installed.packages())[,c("Package", "Version")]'

  • R Studio: available from the JupyterHub home page, clicking on the RStudio button. A new browser window opens in which RStudio can be used, the working directory and files are visible within RStudio

  • Julia: a Julia environment is provided, with a number of libraries installed. You can list the installed packages by running Julia in your terminal with the julia command, then pressing the "]" key and running the status command

To help you get started, we’ve prepared two example notebooks specifically designed for the Machine Learning (GPU) server:

Also you can find them pre-installed on the WEkEO JupyterHub under the following path: ​public/wekeo4data/wekeo-gpu

The JupyterHub environment is fully configured and ready to use, no setup required!

⚠️ The infrastructure includes 5 GPU nodes shared across all users: when all GPUs are occupied, additional sessions will reach a timeout after 15 minutes until a resource becomes available:

Our internal team is working on a fix to avoid such a long timeout and add a clearer message.

Switching between servers


If you want to change the server during your session, follow these simple steps:

  1. Go to File then to the Hub Control Panel

  2. Click on Stop Server

  3. When spawning the JupyterHub again, you can choose another server

What's next?


To know more about the WEkEO JupyterHub, here are some related articles:

Additional resources can be found in our Help Center. Should you require further assistance or wish to provide feedback, feel free to contact us through a chat session available in the bottom right corner of the page.

Did this answer your question?