Overview
This is a provisioning approach that can be used on any type of infrastructure (bare metal, VM, public cloud). This provisioning approach assumes that the lifecycle of the infrastructure (i.e. compute, memory, storage, networking) is managed by the customer. The controller will provision the Kubernetes layer and critical add-ons on the customer provided infrastructure.
Automation Options¶
The below matrix presents a breakdown of actions like creation, upgrading, and deletion of Upstream Clusters across multiple deployment methods: Interactive UI, Declarative RCTL commands, API-driven automation, and Terraform.
| Action | UI | CLI | API | Terraform |
|---|---|---|---|---|
| Create | Yes | Yes | Yes | No |
| Upgrade | Yes | Yes | Yes | No |
| Delete | Yes | Yes | Yes | No |
Bare Metal¶
On bare metal environments, customers are responsible for the lifecycle of underlying Hardware and Operating System. The controller will provision and manage the lifecycle of Kubernetes, the CNI, CSI and the add-ons in cluster blueprints.
Virtual Machines (VMs)¶
On VM based environments, customers have two options:
Bring Your OS¶
Customers are responsible for the lifecycle of underlying Hardware, Hypervisor and the Operating System. The controller will provision and manage the lifecycle of Kubernetes, the CNI, CSI and the add-ons in cluster blueprints.
Pre-Packaged Images¶
Customers are responsible for the ongoing lifecycle of underlying Hardware, Hypervisor and the node Operating Systems.
A current version of the OS (recent patches etc) is deployed during initial provisioning. The customer is expected to implement workflows to keep the OS current and up to date.
The controller will provision and manage the lifecycle of Kubernetes, the CNI, CSI and the add-ons in cluster blueprints. Customers can use the provided pre-packaged images in (a) qcow2 format (typically for OpenStack) or (b) ova format (typically for vSphere) to provision the clusters.
Videos¶
Watch a video of provisioning of a four node, "Multi Master" upstream Kubernetes cluster on "CentOS" with "GlusterFS and Local Storage" as storage classes.
Watch a video of provisioning of a four node, "Multi Master" upstream Kubernetes cluster on "CentOS" with only Local Storage".
Did you know ?
Bare Metal/VM Cluster Lifecycle can be managed through various methods:
Getting Started¶
We have quick start exercise to explore the platform's capabilities for "Lifecycle Management" for Upstream Kubernetes on Bare Metal and VM Environments Bare Metal/VM Clusters.