Deploying and Scaling the Notebook



In [1]:

    
from IPython.display import Image

How could you deploy the Jupyter Notebook in your organization?

Here are the basic building blocks of the Jupyter architecture



In [2]:

    
Image('images/JupyterBlocksA.png')









    Out[2]:



In [3]:

    
Image('images/JupyterBlocksB.png')









    Out[3]:



In [4]:

    
Image('images/JupyterBlocksC.png')









    Out[4]:

JupyterHub is a new Jupyter project that allows a centralized, multiuser deployment of the Jupyter Notebook.

This works by putting individual users's notebooks behind a central proxy and hub.



In [5]:

    
Image('images/JupyterBlocksD.png')









    Out[5]:

This opens up many choices and opportunties:

Users and authentication:
- Unix auth (PAM)
- GitHub OAuth
- Google OAuth
- Active Directory
Store notebooks:
- Google Drive
- OpenStack Swift
- PostgreSQL
- ???
Spawn notebooks/kernels:
- Local subprocesses
- Docker
- Docker+Swarm

The Quantopian Research Platform is an algorithmic/quantitative trading platform for Python using the building blocks of JupyterHub.



In [6]:

    
Image("images/quantopian-research.png", width="90%")









    Out[6]:

Using these same building blocks, we are offering instant, transient, cloud hosted notebooks.

These building blocks are starting to enable ubiquitous, computational narratives

Like Legos, these building blocks can be assembled in a wide range of different ways to suite your usage cases:



In [7]:

    
Image('images/JupyterBlocksE.png')









    Out[7]: