The Jupyter Kernel Gateway makes possible the following novel uses of kernels:
- Attach a local Jupyter Notebook server to a compute cluster in the cloud running near big data (e.g., interactive gateway to Spark)
- Enable a new breed of non-notebook web clients to provision and use kernels (e.g., dashboards using jupyter-js-services)
- Scale kernels independently from clients (e.g., via tmpnb, Binder, or your favorite cluster manager)
- Create microservices from notebooks via
The following diagram shows how you might use
tmpnb to deploy a pool of kernel gateway instances in Docker containers to support on-demand interactive compute:
For more inspiration, see the jupyter/kernel_gateway_demos.