Kyvos Enterprise Deployment Guides
This documentation is for Kyvos 2026.8.x. To view documentation for other Kyvos versions, click here. |
|---|
Overview
This chapter is the starting point for installing and deploying Kyvos across supported environments — AWS, Azure, GCP, and on-premises. Use it to find the right prerequisites, configuration, and deployment guide to successfully set up Kyvos and its components.
Supported Kyvos Deployment Methods
Kyvos supports two ways to deploy a cluster:
Method | Description |
|---|---|
Automated Deployment | Kyvos provisions and configures the required cloud resources for you, using a predefined configuration. Best suited when resources don't already exist and you don't need to specify detailed, custom, per-service configuration. |
Wizard-Based Deployment | You provide configurable inputs and detailed, service-level configuration, and deploy onto resources that already exist (or that need custom naming, tagging, or creation methods Automated deployment doesn't support). |
The right method depends on factors such as whether resources are pre-created, the permissions available, naming and tagging requirements, and the target environment (cloud vs. on-premises). See the platform-specific deployment guides linked below for detailed guidance on choosing between the two.
On this page:
Deployment by Platform
When to Use Automated Deployment vs. Wizard Deployment
Area | Use Automated | Use Wizard |
|---|---|---|
Resource existence | Resources do not already exist and need to be created. | Resources on which the cluster is to be deployed already exist. |
Permissions for resource creation | Sufficient permissions exist to create the required resources. | Sufficient permissions to create the required resources are not available. |
Extent of required permissions | Permissions needed are limited to:
| Permissions needed are:
Note: If List permission is unavailable, resources cannot be auto-populated or selected from a list — users must enter resource details manually. |
Resource naming convention |
| A custom or complex naming convention is required that Automated deployment does not support (pre-create the resources with the required names, then use Wizard). |
Deployment complexity | Deployment is straightforward, with no requirement for custom tagging. | Deployment is complex and requires custom tagging on resources. |
Resource grouping | Resources are expected to be created in the same group/region. | Resources are spread across different groups/regions. |
Nodes | Multi-node deployment is expected. | Applies regardless of whether single-node or multi-node deployment is needed. |
Environment | Cloud environment only. | Cloud or On-Prem. |
Authentication | Key-based. | Key-based or Password-based. |
Memory mode | Auto. | Manual or Auto. |
Cluster mode | Standard. | Standard. |
Compute type |
|
|
Ease of use | User wants a ready-to-go, auto-deployed cluster with predefined configuration, without specifying detailed/custom settings. | User needs to configure the deployment explicitly, including detailed, service-level configuration. |
Supported resource-creation method | Cloud-native resource creation is allowed:
| Required when organizational policy:
|
Image type | Standard (or Custom) image support is sufficient. | Neither the Standard nor Custom image capability of Automated deployment meets the requirement. |
OS & OS version | OS and OS version are supported by Automated deployment:
| OS or OS version is not supported by Automated deployment. |
Kyvos Compute
How Semantic Models are Processed
The compute infrastructure used for semantic model processing depends on the deployment environment and selected compute option:
AWS and Azure: Semantic models can be processed using Kubernetes-based nodes or dedicated compute nodes.
GCP: Semantic models are processed using dedicated compute nodes.
On-premises: Semantic models are processed on dedicated nodes.
Kyvos Native Compute Options
Kyvos Enterprise supports the following Kyvos Native compute options:
Shared Query Engines: Query Engines handle both query execution and semantic model processing.
Dedicated Compute: Semantic model processing runs on dedicated compute nodes, independently of Query Engines.
Kubernetes: Semantic model processing runs on Kubernetes-based compute nodes.
Compute Type Support by Deployment Method
For Automated Deployment
AWS
Kyvos Compute Type | Enterprise | Marketplace | Marketplace Free | SaaS | SaaS SNS |
Dedicated Compute | Yes | Yes | Yes | Yes | Yes |
Shared Query Engines | Yes | No | Yes | No | Yes |
Kubernetes | Yes | Not Applicable | Not Applicable | Not Applicable | Not Applicable |
Azure
Kyvos Compute Type | Enterprise | Marketplace | Marketplace Free | SaaS | SaaS SNS |
Dedicated Compute | Yes | Yes | Yes | Yes | Yes |
Shared Query Engines | Yes | No | Yes | No | Yes |
Kubernetes | Yes | Not Applicable | Not Applicable | Not Applicable | Not Applicable |
GCP
Kyvos Compute Type | Enterprise | Marketplace | Marketplace Free | SaaS | SaaS SNS |
Dedicated Compute | Yes | Yes | No | Yes | Not Applicable |
Shared Query Engines | Yes | Yes | Yes | Yes | Not Applicable |
For Wizard-Based Deployment
AWS
Kyvos Compute Type | Enterprise |
Dedicated Compute | Yes |
Shared Query Engines | Yes |
Kubernetes | Yes |
Azure
Kyvos Compute Type | Enterprise |
Dedicated Compute | Yes |
Shared Query Engines | Yes |
Kubernetes | Yes |
GCP
Kyvos Compute Type | Enterprise |
Dedicated Compute | Yes |
Shared Query Engines | Yes |