A prototype for interactive analysis

This project is a prototype for a data analysis system, CMS compliant. The main targets are:

1. Reducing analysis "time to insight" (training time for newcomers included) with an interactive and user-friendly UI
2. Single and easily accessible hub to reduce the complexity and maintenance of multiple and slightly overlapping solutions
3. Increasing the system delivered throughput (evts/s)

Current design

The environment is composed using:

• JupyterHub and JupyterLab to manage the user interaction part of the infrastructure
• Dask to introduce the scaling over a batch system
• XRootD as data access protocol toward AAA

At the moment, it supports scaling over HTCondor clusters using a custom dask-jobqueue module.