DataLab

Custom JupyterLab container for local-workstations and in-cluster Kubernetes Data Science, Machine Learning and IoT.

DataLab is a large Jupyter Notebook container assembled for use as a central IDE within a data platform. JupyterHub provisions DataLab and uses a Kubernetes Service Account with a well defined RBAC configuration.

Read more about JupyterHub on Kubernetes.

Local Workstation / GPU Station

DataLab is a feature-packed Data Science, Machine Learning and IoT IDE for use in-cluster, or on local workstations. DataLab contains kubefwd, allowing developers and data scientists to work locally with APIs and data sources exposed through Kubernetes services and port-forwarded directly into the JupyterLab environment.

NOTE: kubefwd is included for operating this JpuyterLab container on a local workstation.

docker run --rm --name datalab -p 8888:8888 \
  -v "$(pwd)":"/home/jovyan/work" \
  -v "$HOME/.kube/":"/home/jovyan/.kube/" \
  --user root \
  -e GRANT_SUDO=yes \
  -e JUPYTER_ENABLE_LAB=yes -e RESTARTABLE=yes \
  txn2/datalab:v0.0.6

Packages

Datalab contains the following utilities:

• git
• kubectl
• kubefwd (for local workstations)
• nuctl nuclio command line (Serverless)

Kernels:

• Python 3
• Octave 4.2.2
• Julia 1.1.1
• R 3.6.1

Python Libs:

• nuclio-jupyter
• pylantern
• ipywidgets
• seaborn
• kubernetes
• kafka-python
• elasticsearch
• rubix
• python-gitlab
• jupyterlab-git
• scipy
• numpy
• pandas
• scikit-learn
• matplotlib
• tensorflow
• torch
• torchvision
• fastai
• octave_kernel
• paho-mqtt
• web3
• py-solc

Build

docker build -t txn2/datalab:v0.0.6
 .