What is Jupyter Notebook
Project Jupyter is a non-profit, open-source project, born out of IPython Project in 2014 as it evolved to support interactive data science and scientific computing across all programming languages. Jupyter is 100% open-source software, free for all to use. Jupyter is developed open for use with GitHub, through the consensus of the Jupyter community. The purpose of jupyter is to support interactive data science and scientific computing across all programming languages.
The Jupyter Notebook is an open source web application that you can use to create and share documents that contain live code, equations, visualizations, and text. Jupyter Notebook is maintained by the people at Project Jupyter. Jupyter’s name references the 3 languages: Julia, Python, R.
Jupyter Notebook at University of Maryland
Here at UMD Jupyter Notebooks are used by faculty, staff and students for a variety of reasons for instructional, research and analysis purposes. The primary programming language for using Jupyter Notebook is Python, The use of Jupyter Notebook requires data storage whether it is local hard drive, Network Storage Share, Google Drive, GitHub, and Google Cloud Storage for your datasets and results generated using Jupyter Notebook
At UMD, there are the most frequently used options of Jupyter. Two solutions are cloud based and two are on premise implementations.
- Colaboratory (CoLab) via Google Workspace
- User or Department Local System Implementation
- DIT HPC Open OnDemand
- Google Cloud Platform
All solutions will cost depending on your requirements. The basic configurations of Colaboratory and local user workstation or laptop has no costs associated with it. However if your needs exceed the basic configuration of Colab or your system, you should expect and budget for those additional costs.
Jupyter Notebook options and use cases
Here is a quick overview on your different choices and best use cases for each.
Colaboratory aka CoLab
- Cloud-based Jupyter Notebook environment.
- CoLab lives in Google Workspaces.
- Save work to your UMD Google Drive or GitHub.
- Supports Python 3 and 3rd party tools..
- Access a two core 2.3GHz with 16GB RAM
- Free tiers to GPUs and TPUs.
- Max runtime of 12 hours.
- No configuration required.
- Is accessible via UMD domain.
- Usually used for instructional purposes.
- Based configuration is free to all umd faculty, students and staff.
- Intended for single user implementation.
- Colab Pro or Colab Pro+ not supported in UMD domain not currently available for discounted fee offerings for UMD.
Local system implementation
- Local installed solution.
- Can be installed on workstations, laptop or network server.
- Supported on both Windows and Linux Operating Systems.
- Must installed minimal components for user experience (Python, Numpy, Matplotlib, Jupyter).
- Managed by user.
- Will require network configuration setting for server based implementation.
- Recommended at least two - four core cpu with 16GB-32 GB RAM depending on your requirements.
- Data can be stored locally, across network share, cloud.
- Server provides multiple instances but not shared projects (If you wish a server implementation, contact DIT for Virtual Machine solution).
- Used for single user research and analysis.
UMD High Performance Cluster (HPC) Open OnDemand Jupyter Application
- An interactive jupyter session on the HPC cluster compute nodes with graphics support.
- Single user Instance virtual machine.
- Sessions are default 1 core cpu and runs on RHEL8 OS only.
- GPU configuration available.
- Python and R applications are default options.
- Additional applications can be installed when requested (customized template).
- Fee required for access to this environment using KFS.
- All data is stored in local HPC cluster storage.
- Managed Service by DIT HPC team.
- Used for single user research and analysis (data scientist, researchers).
- Great for those with older hardware.
Google Cloud Platform (GCP)
- Setup within the UMD sponsored GCP environment.
- Has two options:
- Build your own (single jupyter notebook or JupyterHub).
- GCP serverless solution (Vertex AI).
- AI version comes configured with python, R and others.
- Instance runs on Linux Operating System Only.
- Datasets can be accessed over the internet.
- Data results are stored either via GCS, GitHub or local system hard drive.
- Have access to other GCP services like Big Query, AutoML, Dataproc and Dataflow.
- Data is encrypted once uploaded.
- Ideal for data scientists, researchers and departments with multiple.
- Requires umd domain account and umd generated GCP project.
- Discounted Google Cloud Fees required for use (~$.05 to $.30 hourly) for standard use no GPUs.
- System Idle fees will apply.
- Access is managed by DIT Research Computing.
All of these solutions are available for the UMD community. For more information firstname.lastname@example.org or submit a GCP Service Request.