Cluster

Cluster is the infrastructure where containers run. Containers are the deployment units and execution processes of applications. Clusters provide computing resources (CPU, Memory, Storage, Network) necessary for container execution.

A cluster consists of nodes (physical or virtual machines) connected via a network. It is an architecture designed for distributed processing. When containers are deployed to a cluster, they are executed on appropriate nodes. This process, called scheduling, is managed by Kubernetes. Kubernetes is responsible for container scheduling and management within the cluster.

Clusters scale resources by adding nodes. If more resources are needed, nodes are added accordingly, and Kubernetes deploys and manages containers on the expanded nodes.

Container networking and storage for data storage are also components of a cluster.

Kubernetes

Kubernetes is a container orchestration engine that runs containers in clusters and manages their lifecycle. Originally developed by Google, it is now maintained as a CNCF (Cloud Native Computing Foundation) project.

Kubernetes is installed on the cluster and is responsible for managing and providing resources required by containers based on the cluster infrastructure (nodes, network, storage).

Node

A node is one or more compute machines that make up a cluster. They can be physical or virtual machines, each equipped with CPU, Memory, and Disk, connected via a network. Nodes are managed by Kubernetes for scheduling.

Nodes are divided into master nodes and worker nodes. Master nodes host the control plane components of Kubernetes and manage the cluster by communicating with worker nodes.

Worker nodes are where application containers are deployed. The number of worker nodes increases based on the number and capacity of applications. The Kubernetes scheduler on the master node is responsible for deploying containers to worker nodes.

Container Network

Containers running on one or more nodes need to communicate with each other, which is managed by the container network.

Container networking is installed as a Kubernetes component. Kubernetes itself does not provide container networking but offers a standardized interface for external providers to supply plugins, known as the Container Network Interface (CNI). Examples of open-source CNI plugins include Flannel, WeaveNet, Calico.

Cocktail Cloud offers options to configure the cluster's CNI plugin.

Ingress Controller

External access to containers is handled by the ingress controller. It routes external traffic to containers based on hostnames and paths. Routing rules are configured for each application and applied to the ingress controller.

The ingress controller is a Kubernetes component. NGINX controller is commonly used and provided as a default in Kubernetes. Other third-party ingress controllers are also available.

Cocktail Cloud offers options to configure the ingress controller.

Storage

Cluster storage provides persistent volumes for container data storage.

Since containers can be rescheduled to different nodes in case of node failure or resource shortage, storing container data on nodes can be problematic. Therefore, a separate volume called a persistent volume is needed to store and manage data safely.

Kubernetes creates and provides persistent volumes through storage classes. When configuring the cluster, an appropriate storage class for storage must be installed.

Cocktail Cloud provides storage classes as addons, allowing users to select and automatically manage suitable storage classes.

Addon

Besides networking and storage, Kubernetes has components to extend its functionality, referred to as addons.

These addons provide additional capabilities to Kubernetes clusters beyond container management. Examples include monitoring and service meshes.

Cocktail Cloud offers various Kubernetes extension components as addons. They are automatically managed from installation to upgrade, and users can choose and use the required addons.