Jupyter Notebook in Dataiku Projects

[sj0071992]

Hi Team,

I just want to confirm that the Jupyter Notebook used in the Dataiku projects are the AWS Jupyter Notebook?

If yes, then how are we using that in Dataiku?

Is it using Iframe or any other approach?

This is just to understand the architecture and let my team know about the backend architecture.

Thanks in Advance

[Clément_Stenac]

Hi,

DSS comes with its own fully integrated Jupyter server, which is iframed in DSS.

[tgb417]

@Clément_Stenac ,

A related question, I think.  

What is the difference between Jupyter notebooks in V10 and before and the new Code Studios in V11 that support Jupyter labs architecture?  I have a sense that there may be a significant difference in the cases of Code Studios.  Can you help us understand the difference architecturally?

[Ignacio_Toledo]

Hi @tgb417,

I'm also interested in what can @Clément_Stenac share from a more technical perspective. From the high level point of view, there are major differences.

• Jupyter Labs as a Code Studio, requires your instance to be configured to use the elastic AI capabilities (connection to a K8S cluster, for example).
• You create Code Studios by project, is not a single instance of the Jupyter Lab that can then be used by any other project in your instance.
• The Code Studio doesn't connect or interact with your existing notebooks. The main idea, as far as I can get, is that you use the Code Studio to develop recipes and code with a more rich IDE than you get when using the integrated DSS editor.

That is what I've discovered so far.

[tgb417]

@Ignacio_Toledo 

Yeah,  That sounds like some of what I think I've heard as well.  But, I feel like I'm missing a bunch of details.

--Tom