Skip to content

Overview

This is an introductory Get Started guide that will provide users with a gentle introduction users to the capabilities of Rafay's GPU PaaS when used by an organization. This guide can be broken down into the steps described in the image below.

Sequence of Steps

At scale, when supporting 10s or 100s of end users, the underlying infrastructure can span multiple data centers or multiple clouds.

At Scale

Step 1: Provision GPU enabled Infrastructure

The Org Admin would have already cloned the required templates from the Rafay Template Catalog into your project.

In this step, as an Infrastructure Administrator, we will provision a Kubernetes cluster with GPUs attached to it. We will ensure that the cluster is configured with critical software add-ons to ensure it is ready for use in subsequent steps.

Step 2: Publish SKUs

In this step, as a PaaS Administrator in the PaaS Studio portal, you will configure and publish Compute and Service Profiles for end users. Once published, these will be immediately available as self service consumption SKUs for end users that will consume it via the Developer Hub portal.

Step 3: End Users

In this step, as an End User you will use the self-service Developer Hub portal to request for a GPU enabled compute instance and deploy a Jupyter notebook on the newly launched compute instance. You will then invite collaborators to your workspace so that they can work with you.