Architecture Overview

This article describes the overall architecture for Platform9 Self Managed Cloud Platform (SMCP).

Platform9 SMCP consists of three major components that make up the product:

• The Management Server: This is the main element that provides component management

• Host Components: These are cluster elements that make management of the Kubernetes cluster possible.

• Utilities: The CLI that aids in setting up the management server

The illustration below is an approximate overview of the configured architecture. It is divided into two main types of hardware:

• A Management Cluster

• The cluster machines

Management Cluster

One of the key prerequisite is a place to install different utilities and the management plane. A management station or deployment host assumes that responsibility. The management cluster is a machine where the management plane can run and also has the necessary connectivity from which users can access all the Kubernetes nodes. This host can be a small workstation like an Intel (Next Unit of Computing) NUC.

The Management Plane

The Platform9 Management Plane, is a unit of deployment that typically represents a region for a given customer. It provides APIs for managing PMK clusters. The management plane runs as a group of containerized services on a Kubernetes cluster. The Management plane is the component that contains various functions like the following:

• Authentication services

• Cluster management services

• Host management services

• Security and tunneling services

• Data services, including message queues and databases

• Monitoring services like Grafana or Prometheus

Further details can be found on the auto$ page.

Cluster Components

Each management service controls one or more Kubernetes clusters, each cluster contains not only the Kubernetes services but also the services/components needed to help manage the Kubernetes cluster itself and other add-ons. These include:

• Kubernetes components like the api-server, scheduler, etcd, kubelet etc.

• CNIs: Calico, Flannel, SRIOV etc.

• Monitoring: Prometheus

• Supporting services to manage and upgrade the Kubernetes cluster.

Further details on the cluster internals can be found on the auto$ and Cluster Internals page.