Profile Bundles
Profile Bundles are reusable bundles that package infrastructure and application configurations for consistent, repeatable deployments across Compute Pools. Instead of manually configuring each deployment's software components, you define it once in a Profile Bundle and reuse them wherever needed.
Profile Bundles are either created in the PaletteAI user interface (UI) or imported from the Palette AI Studio. We offer a catalog of ready-to-use Profile Bundles that support a wide range of use cases, including AI/ML applications, infrastructure, and fullstack deployments. The Profile Bundle is used in most PaletteAI workflows, such as creating a new Compute Pool or deploying an application. The profile bundle is composed of two key components: a Palette Cluster Profile and a Workload Profile.
Components of a Profile Bundle
Cluster Profiles and Workload Profiles play an important role in Profile Bundles. At a high level, Cluster Profiles establish the cluster's infrastructure, while Workload Profiles handle deploying applications to the cluster.
Cluster Profiles
Cluster Profiles contain the core infrastructure software needed to deploy a Kubernetes cluster, such as the operating system (OS), Kubernetes distribution, Container Network Interface (CNI), and Container Storage Interface (CSI). The Cluster Profile is primarily used for infrastructure-related use cases, such as creating a new Compute Pool. A Profile Bundle may contain multiple Cluster Profiles, but it may only contain one Infrastructure or Full Cluster Profile. Cluster Profiles are created and managed in Palette, typically by a platform engineer or administrator.
Types of Cluster Profiles
There are three types of Cluster Profiles:
-
Infrastructure - Provides the essential components for workload cluster deployments within a Palette project: OS, Kubernetes distribution, network, and storage. Collectively, these layers form the infrastructure for your cluster. For more information, review the Create an Infrastructure Profile guide.
-
Add-on - Composed exclusively of add-on layers. Typically, add-on profiles do not contain infrastructure components and are instead designed for reusability across multiple clusters and multiple projects within a tenant. Since they provide the flexibility to configure clusters based on specific requirements, add-on profiles can be added to infrastructure profiles to create a full profile. For an overview of how to build add-on profiles using various types of layers, review the Create an Add-on Profile guide.
-
Full - Combines infrastructure packs with add-on layers. By adding layers, you can enhance cluster functionality. For example, you might add system apps, authentication, monitoring, ingress, load balancers, and more to your cluster. Refer to the Create a Full Profile guide for more details.
Workload Profiles
Workload Profiles are used for application-related use cases, such as deploying a specific application or multiple applications when using the App Deployment workflow. A Profile Bundle may contain multiple Workload Profiles, but it must contain at least one Workload Profile of the type Application or Model. Workload Profiles are created and managed in PaletteAI; both platform engineers and end users can create and manage them. This is one of the main differences between a Cluster Profile and a Workload Profile. The other key difference is that Workload Profiles are primarily intended for deploying end-user applications, while Cluster Profiles are intended for deploying infrastructure software components.
Types of Workload Profiles
There are three types of Workload Profiles:
-
Application - Used to deploy end-user-facing applications, including their required dependencies that are not infrastructure-related.
-
Model - Used to deploy AI/ML models, including their required dependencies, such as the inference engine, if needed.
-
Infrastructure - Used to deploy infrastructure-related dependencies that are not part of the Cluster Profile. This is useful when you want to deploy an infrastructure-related dependency into the Compute Pool that you do not want Palette to manage or when you have a dependency that resides in a private Helm or OCI registry that is defined in PaletteAI but not in Palette.
The following table summarizes the use cases for Cluster Profiles and Workload Profiles.
| Component | Use Case | Managed in Palette | Managed in PaletteAI | Example |
|---|---|---|---|---|
| Cluster Profile | Infrastructure-related use cases | Yes | No | A Cluster Profile that contains the core infrastructure software that is needed for deploying a Kubernetes cluster, such as the OS, Kubernetes distribution, CNI, and CSI. |
| Workload Profile | Application-related use cases | No | Yes | A Workload Profile that contains the Helm chart and the values.yaml configuration for an application |
The following diagram shows the components of a Profile Bundle, including where they are sourced and managed. Cluster Profiles are created and managed Palette, while Workload Profiles and the Profile Bundle itself are sourced from PaletteAI.
Types of Profile Bundles
There are three types of Profile Bundles:
-
Application - Used to deploy AI/ML applications through the App Deployment workflow. Must be paired with an Infrastructure profile to deploy a Compute Pool. Review the Application Profile Bundle section for more details.
-
Infrastructure - Contains the required infrastructure software components needed to create a Compute Pool. Check out the Infrastructure Profile Bundle section for more details.
-
Fullstack - Contains the required components to deploy an AI/ML application and a dedicated Compute Pool at the same time. To learn more, check out the Fullstack Profile Bundle section.
Supported Workflows
The following table illustrates the different types of Profile Bundles and the common workflows that you can use them in.
| Workflow | Compute Pool Type | Profile Bundle Type |
|---|---|---|
| App Deployment | Dedicated | Fullstack |
| App Deployment | Shared | Application |
| Compute Pool Deployment | Shared | Infrastructure |
| Compute Pool Deployment | Dedicated | Infrastructure |
There are more options available when using Profile Bundles in the App Deployment and Compute Pool Deployment workflows. Expand the details below to learn more about mixing and matching Profile Bundle types.
Mixing and Matching Profile Bundle Types
There are scenarios where you can mix and match the Profile Bundle types, as long as the core workflowrequirements are met. For example, a Compute Pool deployment requires Infrastructure software components.
| Workflow | Compute Pool Type | Supported Profile Bundle Options | Notes |
|---|---|---|---|
| App Deployment | Dedicated | Infrastructure + Application | Together, the two profile bundles meet the requirements for deploying an AI/ML application to a dedicated Compute Pool. |
| App Deployment | Shared | Fullstack | The Fullstack profile bundle meets the requirements for deploying an AI/ML application to a shared Compute Pool. Since the Compute Pool already exists, the infrastructure content will be ignored; only the application will be deployed. |
| Compute Pool Deployment | Dedicated | Fullstack | The Fullstack profile bundle meets the requirements for deploying a dedicated Compute Pool. The Workload Profile of the type Application is ignored; only the infrastructure content will be deployed. |
| Compute Pool Deployment | Shared | Fullstack | The Fullstack profile bundle meets the requirements for deploying a shared Compute Pool. The Workload Profile of the type Application is ignored; only the infrastructure content will be deployed. |
Application Profile Bundle
Application Profile Bundles contain only Workload Profiles of the type Application. You can use the Workload Profile Playground editor to create a new Workload Profile that can later be added to the Application Profile Bundle. Application Profile Bundles are used in the App Deployment workflow to deploy an AI/ML application to an existing Compute Pool. For scenarios where you want to deploy an AI/ML application to a new Compute Pool, use a Fullstack Profile Bundle instead.
The following table summarizes the purpose of and illustrates an example use case for Application Profile Bundles:
| Use Case | Profile Bundle Type | Contains Workload Profiles | Contains Cluster Profiles | Example |
|---|---|---|---|---|
| App Deployment | Application | Yes | No | A data scientist or ML practitioner wants to deploy ClearML onto a shared Compute Pool that was prepared by a platform engineer. |
Infrastructure Profile Bundle
Infrastructure Profile Bundles are used in the Compute Pool Deployment workflow to deploy a new Compute Pool. They contain the OS, Kubernetes distribution and version, CNI, and CSI. Other infrastructure-related software components may be included, or other software deemed necessary, before applications are deployed into the Compute Pool.
Infrastructure Profile Bundles are designed for infrastructure use cases. By leveraging Cluster Profiles, PaletteAI can stand up the Compute Pool to support AI/ML applications deployed through the App Deployment workflow.
Platform engineers are the expected users of Infrastructure Profile Bundles, as they are the most likely persona to manage infrastructure resources.
An Infrastructure Profile bundle must contain at least one Cluster Profile of the type Infrastructure or Full. You can include other Add-on Cluster Profiles to add other required infrastructure software components, such as monitoring, ingress, load balancers, and more.
Infrastructure Profile bundles typically contain Cluster Profiles only. However, you may include a Workload Profile of the type Infrastructure if you need to deploy a dependency outside a Cluster Profile. Using a Workload Profile of type Infrastructure is useful when you want to deploy an infrastructure-related dependency into the Compute Pool that you do not want Palette to manage or when you have a dependency that resides in a private Helm or OCI registry that is defined in PaletteAI but not in Palette.
The following table summarizes the purpose of and illustrates an example use case for Infrastructure Profile Bundles:
| Use Case | Profile Bundle Type | Contains Workload Profiles | Contains Cluster Profiles | Example |
|---|---|---|---|---|
| Compute Pool Deployment | Infrastructure | May contain Workload Profiles of the type Infrastructure | Yes | A platform engineer is deploying a shared Compute Pool that is intended for a team of data scientists or ML practitioners to use. |
Fullstack Profile Bundle
Fullstack Profile Bundles are for App Deployment workflows that include infrastructure provisioning. For example, a use case where a data scientist or ML practitioner wants to deploy an application and the required infrastructure simultaneously.
Fullstack Profile Bundles contain a Cluster Profile of the type Infrastructure or Full, as well as a Workload Profile of the type Application. Think of the Fullstack Profile Bundle as a combination of the Infrastructure Profile Bundle and Application Profile Bundle, combined into one Profile Bundle.
In an App Deployment, you may use a Fullstack Profile Bundle on a shared Compute Pool. However, in this case, PaletteAI ignores the Cluster Profile and deploys the Application only through the Workload Profile of type Application. In other words, since the Compute Pool already exists, the infrastructure content will be ignored; only the application will be deployed. The same applies to a Compute Pool Deployment, you may use a Fullstack Profile Bundle to deploy a Compute Pool, but the Workload Profile of the type Application will be ignored; only the infrastructure content will be deployed.
The following table summarizes the purpose of and illustrates an example use case for Fullstack Profile Bundles:
| Use Case | Profile Bundle Type | Contains Workload Profiles | Contains Cluster Profiles | Example |
|---|---|---|---|---|
| App Deployment | Fullstack | Yes | Yes | A data scientist or ML practitioner wants to deploy run:ai and get the required infrastructure along with a specific OS and Kubernetes distribution and version. The platform engineer does not need to be involved in the deployment, as all the required infrastructure software components are already defined in the Profile Bundle. |
| Compute Pool Deployment | Fullstack | Yes | Yes | A platform engineer wants to deploy a Compute Pool that is intended for a team of data scientists or ML practitioners to use. An existing Fullstack Profile Bundle contains all the required infrastructure software components. In this scenario, the Workload Profile of the type Application is ignored; only the infrastructure content will be deployed. |
A Fullstack Profile Bundle must contain at least one Cluster Profile of the type Infrastructure or Full and one Workload Profile of the type Application. The Profile Bundle can include additional Cluster Profiles of the type Add-on and additional Workload Profiles of the type Application.
Palette AI Studio
Palette AI Studio is a platform that contains a catalog of ready-to-use Profile Bundles that support a wide range of use cases. You can browse the catalog to find a Profile Bundle that meets your needs, and import it into PaletteAI. Profile Bundles are downloaded as a ZIP file containing all the required files to create a Profile Bundle in PaletteAI. These required files include Cluster Profiles, Workload Profiles, custom definitions, and Flux resources for Helm and OCI repositories.
Profile Bundles in Palette AI Studio are self-contained, meaning they depend on no other Profile Bundles. However, Profile Bundles of the types Fullstack and Infrastructure may depend on packs that your Palette environment may not have. In this case, you must ensure the Pack registry is available in your Palette project.