Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[version-3-4] docs: DOC-1166 image update (#2773) #2836

Merged
merged 1 commit into from
May 14, 2024
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
86 changes: 49 additions & 37 deletions docs/docs-content/architecture/orchestration-spectrocloud.md
Original file line number Diff line number Diff line change
@@ -1,47 +1,57 @@
---
sidebar_label: "Order of Operations"
title: "Order of Operations"
description: "The methods of workload cluster provisioning for K8S clusters with Palette"
description:
"Learn about the order of operations that make up the workload cluster provisioning process in Palette and how Cluster
API is used."
icon: ""
hide_table_of_contents: false
sidebar_position: 10
---

Spectro Cloud Palette provisions standard, upstream Kubernetes clusters using
[Cluster API](https://cluster-api.sigs.k8s.io/).
Palette provisions standard upstream Kubernetes clusters using [Cluster API](https://cluster-api.sigs.k8s.io/). Cluster
API is a Kubernetes sub-project focused on providing declarative APIs and tooling to simplify provisioning, upgrading,
and operating multiple Kubernetes clusters.

Cluster API is a Kubernetes sub-project focused on providing declarative APIs and tooling to simplify provisioning,
upgrading, and operating multiple Kubernetes clusters.
:::info

Check out the [Cluster API concepts](https://cluster-api.sigs.k8s.io/user/concepts) page for a detailed explanation of
the components and concepts used in the Cluster API. Cluster API is a Kubernetes sub-project and different from the
[Kubernetes API](https://kubernetes.io/docs/concepts/overview/kubernetes-api/) that you use to interact with your
Kubernetes clusters.

Cluster API demonstratively manages the lifecycle of a cluster (creation, scaling, enhancement, and deletion) and helps
in automating the process of cluster lifecycle management for platform operations. Cluster API also helps in consistent
and repeated cluster deployment across multiple infrastructure environments.
:::

<br />
Cluster API automates cluster lifecycle management for platform operations. It ensures a consistent experience in
cluster deployment across different infrastructure environments.

## Workload Cluster Provisioning

The following diagram illustrates the order of operations that make up the workload cluster provisioning process.

![workload_cluster_provisioning.webp](/architecture_orchestartion-spectrocloud_provision-flow.webp)

1. New K8S cluster request from user submitted to the cluster management system.
1. A user submits a request to Palette for a new Kubernetes cluster. The request may come from the UI, API, or
Infrastructure as Code (IaC) tools such as Terraform or Crossplane.

2. Palette creates the Cluster-API (CAPI) custom-resource specifications for the target cloud, e.g in VMware this would
translate to: Cluster, vSphereCluster, KubeadmControlPlane (with replica 1), and VSphereMachineTemplate. These
resources are created in the management cluster.
2. Palette creates Cluster-API (CAPI) custom-resource specifications for the target cloud, such as AWS, GCP, etc. For
example, in VMware, this would translate to the following resources getting created: Cluster, vSphereCluster,
KubeadmControlPlane, and VSphereMachineTemplate. These resources are created in the Palette internal management
plane.

3. Cluster API and corresponding cluster-api provider, e.g: cluster-api-provider-vsphere, provisions the first
control-plane node CP-A on the target cloud.
3. Cluster API and the respective cluster-api provider, specific to the target platform, such as
cluster-api-provider-vsphere, provisions the first control-plane node (CP-A) on the target cloud.

4. When CP-A is operational, the management platform will install a Palette agent into the workload cluster and then
perform a pivot of the Cluster API resources.
4. When CP-A is operational, the Palette management platform will install a Palette agent into the workload cluster and
then perform a pivot of Cluster API resources. This means that Cluster API resources and their responsibilities are
now handled by the workload cluster, not the Palette management plane.

5. CP-A agent will retrieve the latest specifications for the cluster, e.g: 3 control-plane, and 3 workers. CP-A will
generate and update the remaining CAPI resources, e.g: update replicas to 3 for KubeadmControlPlane, create the
worker's MachineDeployment or VSphereMachineTemplate. Cluster API running in CP-A will provision the remaining
control plane and worker nodes.
5. CP-A agent will retrieve the required specifications for the cluster, such as the number of nodes. CP-A will generate
and update the remaining CAPI resources, e.g: update replicas to 3 for KubeadmControlPlane and create the worker's
MachineDeployment or VSphereMachineTemplate. The Cluster API active in CP-A will provision the remaining control
plane and worker nodes.

6. The Palette agent will install all the additional add-ons as specified by the cluster's cluster profile (e.g:
logging, monitoring, security).
6. The Palette agent will install all the additional add-ons as the cluster profile specifies.

:::info

Expand All @@ -52,26 +62,28 @@ clusters.

## Why Palette Pivots?

Palette's decentralized model is based on a "decentralized management - local policy enforcement" scalable architecture.
Palette's decentralized model is based on a "decentralized management" design with local policy enforcement that allows
for a scalable architecture.

![distributed_orchestration.webp](/architecture_orchestartion-spectrocloud_distributed-flow.webp)

As part of the workload K8s cluster provisioning, only the first control-plane node is launched by Cluster API, running
in the Palette management cluster. Once the control-plane node is operational, Cluster API resources are _pivoted_ from
the management platform into the target workload cluster.
As part of the workload for the Kubernetes cluster provisioning, only the first control-plane node is launched by the
Cluster API active in the Palette management cluster. Once the control plane node is operational, Cluster API resources
are _pivoted_ from the Palette management platform into the target workload cluster.

The target workload cluster is then responsible for provisioning and maintaining the remaining control-plane and worker
nodes. All Day-2 operations which result in node changes, including OS/K8s upgrades, scaling, and K8s certificate
rotation, are triggered by changes to the Cluster API resources in the target workload cluster.
The target workload cluster is responsible for provisioning and maintaining the remaining control plane and worker
nodes. All Day-2 operations, which may result in node changes, including the operating system and Kubernetes upgrades,
node scaling, and Kubernetes SSL certificate rotation, are triggered by changes to Cluster API resources in the target
workload cluster.

Palette pivots these clusters for several reasons, related to scalability and availability:
Palette pivots Cluster API resources from the Palette management platform to the workload clusters for reasons such as:

- **Scalability** - The management platform scales to meet the demand of all your workload clusters as the number of
tenant clusters and nodes increases in size.
- **Scalability** - The central management platform, Palette, can scale to meet the demand of all your workload clusters
as the number of workload clusters and nodes increase in size. The workload cluster manages its own lifecycle and
resources through guidance from the Palette management platform.

- **Resiliency** - Even if the management platform were to experience an outage, the workload clusters would retain
their resiliency capabilities, auto-recovery, launching of new nodes on failures, auto-scaling, and other policies
still work!
- **Resiliency** - If the management platform were to experience an outage, the workload clusters would retain their
capabilities, such as auto-recovery, launching of new nodes on failures, auto-scaling, and other features.

- **Intermittent network resiliency** - The design supports use cases where the workload clusters can still operate in
- **Network resiliency** - The decentralized design supports use cases where the workload clusters can still operate in
intermittent and disconnected network availability situations.
Binary file not shown.
Loading