Skip to content

Mar

v3.2 Update 1 - SaaS

05 Mar, 2025

The section below outlines enhancements introduced in this update.

GPU Platform as a Service (GPU PaaS)

Clone Profiles

Administrators can now clone compute and service profiles. This substantially reduces the administrative burden associated with creation of new compute and service profiles. Typical use cases are:

  • Clone to same project e.g. admin wants to create a new profile based on an existing profile
  • Clone to a different project e.g. admin wants to onboard a new user or team in a new project

Clone Profile

Note

Please ensure that the environment template associated with the compute/service profile is also shared/available in the new project.

List Service Instances on Compute Instances

Service instances can be launched into compute instances. Users may not know which service instances have been deployed on a specific compute instance. With this enhancement, compute instances listed in the PaaS Studio and the end user facing Developer Hub now display the service instances deployed on it. For example, in the image below, the end user can immediately tell that a service called "qa-notebook" has been deployed to the compute instance "small-qa".

Service Instances on Compute Instance

Global Settings for Providers

Service Providers (e.g. NCPs) and Enterprises providing GPU Cloud or org-wide PaaS can use the Rafay Operations Console to centrally manage settings for compute and service profiles that are shared with downstream customer's Orgs (i.e. tenants). These settings can be configured to apply to "ALL" or "SPECIFIC compute and service profiles. For example,

Centralized Agents

Associate "agents" for workflow execution in compute and service profiles. When end users in downstream orgs launch compute and service instances, the associated environment template will ALWAYS utilize the agent configured under global settings for execution of the workflow.

Soft Limit Enforcement for Compute Profiles

Centrally manage "soft limits" (aka quotas) for compute profiles to help enforce "max compute instances" for a specific compute profile that can be deployed in a customer Org. For example, the provider may wish to enforce "soft limits" for the max number of compute instances for a specific compute profile.

  • Specify a max-instance quota of "5" for the "small" compute profile
  • Downstream Orgs "Apple" and "Mango" can now each launch and operate a maximum of "5" compute instances of type "small"
  • Users will be shown an error message enforcing the quota/soft limit if they attempt to launch a 6th compute instance.

Org Specific Settings

Administrators can use the Rafay Operations Console to centrally manage settings for compute and service profiles for specific tenant orgs. This allows them to provide a customized experience for high value enterprise customers.

Usage APIs for Billing

Fine grained APIs are now available for service providers to programmatically retrieve data to perform billing related calculations.

  • 1. Instance Usage for All Orgs This API provides aggregate usage per instance and the respective billable parameters across all the Orgs.

  • 2. Instance Usage API for Org This API provides aggregate usage per instance and the respective billable parameters for a given Org.

  • 3. Profile Usage for All Orgs This API provides aggregate usage per profile (i.e. SKU) and the respective billable parameters across all the Orgs.

  • 4. Profile Usage for Org This API provides aggregate usage per profile (i.e. SKU) and the respective billable parameters for a given Org.

Info

Navigate to Rafay's Open API Explorer to explore the new Billing APIs