Select the Gcore Platform

Gcore Edge Solutions
Go to Gcore Platform →
  • Edge Delivery (CDN)
  • DNS with failover
  • Virtual Machines
  • Bare Metal
  • Cloud Load Balancers
  • Managed Kubernetes
  • AI Infrastructure
  • Edge Security (DDOS+WAF)
  • FaaS
  • Streaming
  • Object Storage
  • ImageStack (Optimize and Resize)
  • Edge Compute (Coming soon)
Gcore Hosting
Go to Gcore Hosting →
  • VPS Hosting
  • Dedicated Servers
  • SSL Certificates


What Is a Kubernetes Cluster?

Containerization, despite its many advantages, is a challenging software development model. Things can get complicated easily, especially when the containers are deployed at scale. Kubernetes is a system designed to automate the deployment, scaling, and management of containerized applications. It helps to reduce the woes of container management, but its help is only as potent as the health of its components. A Kubernetes cluster is one such component that is critical to the smooth functioning of the Kubernetes system. This article explains what a Kubernetes cluster is and how it works, and explores its use cases and technical benefits.

Understanding Kubernetes Clusters

A Kubernetes cluster is a set of machines, known as nodes, that run containerized applications deployed and orchestrated by Kubernetes. A cluster consists of at least one master node and several worker nodes, which work together to run and scale containerized applications. Usually, multiple master nodes exist to ensure high availability and fault tolerance.

When you deploy your application on a Kubernetes cluster, you provide a set of instructions on how the application should run, how many instances should be running, which nodes they should run on, what resources they should use, and how they should react to different problems. This set of instructions is given to the Kubernetes system using a declarative API, which means that you “declare” what you want to happen, and it’s Kubernetes’ job to make that happen. The master node in the cluster uses this set of instructions to schedule and run containers on the appropriate worker nodes. It also continually monitors the state of the cluster to ensure it matches your declared instructions. If a node or application fails, Kubernetes automatically handles the rescheduling and redeployment based on your declared state, providing self-healing capabilities. These features, alongside its fault tolerance, result in excellent reliability and uptime, leading to a positive end-user experience.

To understand the make-up of a Kubernetes cluster and how it engineers and orchestrates deployed containers, let’s look at Kubernetes architecture more broadly.

The Architecture of the Kubernetes Cluster

The architecture of a Kubernetes cluster comprises a network of interrelated components. that serve interrelated functions that help ensure that the cluster works as expected. Each component has prespecified resources, limits, and values. If workloads are not assigned with respect to these limits and values, the values compound and hamper the smooth functioning of the Kubernetes ecosystem. Some components undertake primary tasks, handling substantial workloads, while others are add-ons used to improve functionality and efficiency.

The below image summarizes the relationship between the core components of the Kubernetes cluster architecture.

Architecture of the Kubernetes cluster showing the components of the control plane and worker nodes
Figure 1: Kubernetes cluster architecture

Let’s now look closely at the components and subcomponents of the Kubernetes cluster architecture displayed above.

Master Node/Control Plane

The master node, also called the control plane, is the central system that executes the overall responsibilities of the Kubernetes cluster. It manages and maintains communication between the worker nodes in the Kubernetes system. The master node schedules workloads and manages the lifecycle of pods and other resources.

Every pod (or workload) is assigned to a worker node with the optimum resources and environment to engineer the pod’s smooth functioning. The kube-scheduler, one of the master node’s components, performs this scheduling task that prevents overhead and ensures the seamless functioning of the Kubernetes ecosystem. See the table below for the four critical components of the master node.

Kubernetes API serverThis is the entrance point for communication between the Kubernetes cluster and external clients. The Kubernetes API server is the primary endpoint for all interactions with the cluster. As such, all tasks—such as creating, updating, and deleting workloads—are executed through it.
kube-controller-managerThe kube-controller-manager is responsible for managing the core components of the Kubernetes cluster, monitoring the cluster’s state, handling API calls, and other workload management roles such as ensuring the maintenance of the desired state of deployments and replica sets.
kube-schedulerThe kube-scheduler interacts with the Kubernetes API server to schedule tasks on worker nodes and ensure optimal resource usage. It assigns pods to nodes based on resource availability, and other constraints such as affinity, anti-affinity, and location.
etcdThis is a distributed key-value store that stores cluster-wide configuration data such as Kubernetes objects and metadata. It can run on multiple nodes in a Kubernetes cluster and ensures data consistency by propagating data updates quickly and consistently across the cluster.

Worker Nodes

Worker nodes are the fundamental building blocks of the Kubernetes cluster. They execute application workloads and run containers. Worker nodes communicate with the Kubernetes master node to get the required service definitions and schedules, trigger pod creations, and report the status of the running containers. Each worker node runs a Kubernetes runtime environment for containerized workloads, such as Docker or CRI-O. The components of a worker node are presented in the table below.

kubeletKubelets run on each worker node. They communicate with the master node to retrieve instructions and schedule container tasks.
kube-proxy  This is a network proxy that runs on each worker node. It manages network communication within the worker node, ensures load balancing of network traffic between containers, listens for Kubernetes services and sets up the necessary rules to route traffic to the appropriate pods.
Container runtimeSoftware responsible for running and managing containers on each worker node. Examples of container runtimes include Docker, CRI-O, and Containerd.
Pod networkingEach worker node must have a network interface that is used to communicate with other nodes in the cluster. Pod networking allows containers running on different worker nodes to communicate with each other as if they are running on the same node.

From the above, we can infer that there is a strong dependency between the master and worker nodes. The image below further shows this relationship.

The relationship between the master and worker nodes in the Kubernetes cluster
Figure 2: Master-worker node relationship


A Pod is the smallest deployable unit in a Kubernetes cluster. One or more containers can be deployed in a pod, such that they share the same network and storage resources. A pod is managed and scheduled to run on a worker node.


A Service is a Kubernetes object that offers a standard method to access one or more pods. It provides pods with stable IP addresses and DNS names. Services expose pods to other parts of the Kubernetes cluster or to external networks, while abstracting details of individual pods, such as their IP addresses. The Service object can also provide several other features, such as session affinity, which allows clients to maintain connections with the same pod over multiple requests.


A Deployment is an object used to manage the creation and scaling of pods. It ensures that the right number of replica sets are running.


Namespaces are a virtual cluster that partitions resources within a Kubernetes cluster. Each namespace is isolated from others and can be used to organize and secure Kubernetes objects.


Volumes are directory mounted to a container in a pod. A volume stores data separately from the container in which it runs. It provides a way for containers to access, share, transfer and persist data beyond the lifetime of the container. Since containers are ephemeral, volume helps to retain data once they stop running.


A ConfigMap is a Kubernetes object used to store configuration data as key-value pairs that can be accessed by applications at runtime. It is used to decouple configuration data from the application code. ConfigMap can be created using YAML files, command-line tools, or the Kubernetes API.


Add-ons are optional components installed on worker nodes that provide additional functionality to the Kubernetes cluster. They serve various purposes, such as monitoring container performance and managing application logs. Examples of add-ons include the Kubernetes Dashboard, Metrics Server, and DNS.


An Ingress manages the cluster’s traffic and routes external requests to the appropriate container. It can route traffic based on hostnames, path, URL schemes, and other conditions. It acts as a Layer 7 load balancer and is used to expose HTTP and HTTPS routes from outside the cluster to services within the cluster. In other words, an Ingress allows external users to access services running inside the Kubernetes cluster, providing a single point of entry to the services.


In a Kubernetes cluster, a DNS is a service responsible for resolving the namespace of Kubernetes objects. It is an add-on that provides a naming system for the containers, pods and services running within the cluster. Additionally, a DNS service allows developers to use standard DNS queries to discover and connect to services running within the cluster. When a pod or service is created, a corresponding DNS record is generated.

Container Network Interface (CNI)

The CNI is a Kubernetes networking standard responsible for providing a uniform networking layer for the cluster. It defines how container runtimes can connect to various network providers. CNI defines a standardized way for Kubernetes networking plugins to handle tasks such as IPAM (IP address management,) network namespaces, and container network interface configuration. It allows Kubernetes administrators to choose the most appropriate networking solution for their specific use case.

How a Kubernetes Cluster Works

A Kubernetes cluster works by automating the deployment, scaling and management of containerized applications. This is further explained in the flowchart below.

A flowchart showing the process by which a Kubernetes cluster works

Managed vs Unmanaged Kubernetes

Kubernetes can be offered as a managed service. Managed Kubernetes is a fully-administered Kubernetes service that provides users with a Kubernetes cluster without them having to manage the underlying infrastructure. This is beneficial because of the complexity of Kubernetes clusters.

Managed Kubernetes services are offered by cloud service providers, including Gcore, who take over the responsibility for setting up, configuring, and operating the Kubernetes clusters. This enables teams to deploy large sets of containers easily, providing scalable, fault-tolerant, and highly-available platforms for deploying and running containerized applications.

With managed Kubernetes, the provider manages the master node and other complex aspects of the infrastructure such as the control plane, its API server, as well as the etcd, scheduler and controller manager. The worker node and computer infrastructure are provided, customized to your needs, and autoscaled (by the provider) according to your configuration settings. When necessary, you can access the worker node via SSH.

Managed Kubernetes can offer significant control over the worker nodes and promises high SLA with the managed master nodes, depending on your choice provider and your organization’s needs. Whatever the extent of management, managed Kubernetes ensures that your containers function at optimal efficiency.

Unmanaged Kubernetes

“Unmanaged” Kubernetes means that you must install, manage, and maintain the Kubernetes yourself. From the installation of the nodes, software packages and all necessary infrastructure, to the management and synchronization of the master and worker nodes to determine on which node the application is run, all the complex procedures and decisions must be made by you.

Benefits of Managed Kubernetes

Some differential advantages of managed Kubernetes include its cost effectiveness, automated scalability, security and access control, and ease of deployment and configuration.

Cost Efficiency

Managed Kubernetes reduces operational overhead as you do not need to invest in hardware infrastructure and human expertise required to manage the master nodes.

Automated Scalability

Managed Kubernetes service providers usually offer automated scaling of nodes to accommodate workload spikes, including Gcore.

Simplified and Quicker Deployment

Since master node management is handled by the service provider, all patches you deploy to the worker node are rolled up seamlessly, which eases and quickens software release.

Security and Access Control

Managed Kubernetes providers offer end-to-end cluster-wide and pod-level security in their platforms, alongside automated security features such as regular up-to-date security patches and bug fixes. By using a managed Kubernetes solution, you can fine tune cluster access to grant root access only to authorized entities.

Dynamic and Speedy Configuration

Managed Kubernetes allows developers to manage application configurations speedily on the fly. This makes it possible to update an application configuration quickly and without redeploying the whole application.

Benefits of Having Root Access to Managed Kubernetes

Root access grants administrative access to the worker nodes. It requires permission and authentication from the provider, and is usually enabled via the SSH protocol by providing appropriate access credentials. Root access is often a feature of managed Kubernetes, and something to look out for when selecting a managed Kubernetes provider.

Root access enables you to grant timed access to worker nodes for troubleshooting purposes. You can also optimize the Kubernetes clusters to meet specific performance and security requirements. However, it is important to note that having this access can pose security risks if not handled appropriately. Exercise caution by only sharing root access with trusted staff and providers on a need-to-use basis.

Kubernetes Clusters Use Cases

Let’s explore some use cases for Kubernetes clusters.

Deploying Microservice-Based Applications

A Kubernetes cluster is designed to deploy and manage microservices-based applications. Kubernetes orchestrates microservices into coherent applications, simplifying their management and scaling while ensuring high availability.

Running Machine Learning Workloads

The Kubernetes cluster also enables the deployment and management of machine learning models, training processes, and model states. This ensures that developers can scale their training workloads on distributed computing resources and manage the post-training serving of their models efficiently.

Managing Internet of Things (IoT) Applications

IoT applications usually have diverse endpoints with different connectivity options. The Kubernetes cluster enables developers to deploy IoT applications, ensuring that devices and sensors run the application logic and serverless functions, thus minimizing data transfers to the central cloud.

Running Continuous Integration and Continuous Deployment (CI/CD)

With the Kubernetes cluster, developers can automate their software deployment pipeline, enabling faster release cycles, better infrastructure utilization, and improved quality control.


The Kubernetes cluster is a vital tool for modern application development, providing a streamlined and efficient way to automate the deployment, scaling, and management of containerized applications. It enables technical decision-makers to manage complex containerized applications at scale across multiple infrastructures.

Gcore offers Managed Kubernetes for businesses and technical decision makers seeking to enjoy the benefits of Kubernetes without the attendant complexities and cost escalations of its unmanaged equivalents.

Subscribe to a useful newsletter

Favorable offers and important news once a month. No spam.