What is Kubernetes? – An Overview

Mati Lerner
Kubernetes
17 Dec 2020

Introduction — What is Kubernetes?

Do I need Kubernetes, what is Kubernetes, and Kubernetes alternatives are all common questions during this resurgence of decentralized computing. The growth of cloud infrastructure has accelerated our push towards centralized computing, but Kubernetes, developed by Google in 2014 as a way to break down application instances to run them closer to users, has shifted our focus back towards the edge.

The origin of Kubernetes meaning lies within Google, as it was a tool they developed to manage application instances for their growing cloud deployments as part of Project Borg and was later given the name Kubernetes – an ancient Greek word that means helmsman. In 2015 open source project Kubernetes was donated by Google to the Cloud Native Computing Foundation. It has since become one of the fastest-growing open-source projects in history. It’s used by industries all over the world and quickly becoming the “de facto” choice for cloud-native applications.

However, businesses overwhelmed with complex application environments are still asking themselves, “do I need Kubernetes?” It’s a tough question, but there are two things to keep in mind: Kubernetes is free, and it integrates with almost every major cloud platform in the world.

With Kubernetes, you can cluster together groups of hosts running Linux containers across individual clouds or a multi-cloud platform. Specifically, applications that require on-demand, rapid scaling. It eliminates the need for many of the manual processes of container workload management and optimization.

How does Kubernetes work?

What does Kubernetes do? Kubernetes (K8s) is an open-source system for automating deployment, scaling, and management of containerized applications. It can integrate with a vast number of existing hardware and software solutions on-premise or in the cloud, or, in some cases, be adapted to a company’s proprietary solution. The smaller scale of containers makes it ideal for deploying updates and testing performance in pre-production environments.

However, complex applications often use multiple containers, especially for applications built using microservices, as these containers are grouped into clusters depending on location and platform. Without the proper tools, this could require manual scheduling for deployments, restarts, resource allocation, and managing networking between each container.

Herein lies the value of automation. Kubernetes handles all of this automatically to ensure that each container is running optimally. It is used to automate the deployment and scaling of clusters across your entire infrastructure, handle resource allocation under heavy loads, and ensure application failures don’t impact the wider organization.

The Kubernetes basics are fairly straightforward, with some niche use cases dependent on existing hardware & software.

Notable features of Kubernetes

Kubernetes can be used in many environments. It also integrates with a variety of third-party and open-source software. The characteristics of Kubernetes are useful to take full advantage of the modern IT landscape. The paragraphs below list a few common ones.

Streamlining Deployment

With Kubernetes IT teams can deploy updates directly to Clusters, often improving deployment timelines and reducing code conflicts in production environments. Depending on the business there is some value in using Kubernetes in pre-production, but for most businesses, the efficiency provided by Containers during application deployments saves a significant amount of time and enables IT operations to maximize hardware resource usage for applications in any environment.

Auto-scaling

Container deployments allow developers to take a hands-off approach to manual application management. Kubernetes will analyze the performance of containers and scale resources to match demand. If new code is deployed or a major spike in traffic suddenly arises, it can adjust the allocated resources as needed thus ensuring there are no drops in performance.

Auto-healing

Kubernetes monitors clusters and containers 24/7 to ensure they’re fully operational. In the event of a malfunction or failure, the system will attempt to reach a stable state again by restarting or spinning down non-responsive containers. The auto-healing mechanism will then create and provision another node to replace the unhealthy one, thereby ensuring the entire system remains stable.

Cloud Portability

Using Kubernetes enables organizations to build applications in one place and deploy them on any infrastructure. The apps built are not dependent on a specific environment. As a result, organizations can take advantage of public cloud, private cloud, multi-cloud, and hybrid cloud environments.

Kubernetes Architecture

But how does Kubernetes fit into the bigger picture of the IT ecosystem?  In this section we will introduce how each of the components of Kubernetes fit together as small pieces of a larger puzzle. In the image below you can see the basic structure of a Kubernetes Cluster.

A Kubernetes Cluster Illustration

For more information regarding the meaning of each term, please refer to the terminology section. 

What is a Kubernetes Cluster?

Each cluster is built from a number of smaller parts, starting with containers. Containers and Kubernetes are closely associated, and are most likely the term you hear most often when learning Kubernetes, as they contain the actual application code running on your infrastructure.

The containers are contained in pods, which are grouped into nodes – virtual or bare metal. These Pods and Nodes are managed by a Kubernetes Control Plane where IT teams can monitor and assign parameters when installing Kubernetes.

The Control Plane is responsible for global scheduling, new pod assignment, and responding to cluster events. When new Pods are deployed, the Scheduler decides where those Pods are  assigned based on current availability. If a Pod fails, the Control Plane is responsible for restarting it to ensure performance. An Ingress Kubernetes controller is responsible for exposing the HTTP and HTTPS routes outside the controller to services within the cluster, enabling the Control Plane to handle routing for external services.

The final component of the Kubernetes Control Plane that exposes the Kubernetes API is the kube-apiserver. It exposes the Kubernetes API, which acts as the front end for the Kubernetes Control Plane. Using it, developers can modify parameters and monitor the performance of clusters.

Please note that the kube-apiserver scales horizontally. As a result, scaling it will deploy more instances. Multiple kube-apiservers can also be run to balance traffic between instances.

Advantages of Kubernetes

If you’re up to this point, you may have noticed some of the benefits of using the Kubernetes platform in your IT environment. Below, you will find a additional advantages to using Kubernetes, especially on the DevOps side and with uniform tooling for heterogeneous environments (private, public clouds):

  • Kubernetes orchestration across multiple hosts
  • Scaling containerized applications and resources on-demand
  • Faster application deployments and updates with full control
  • Full automation with self-healing, auto-placement, auto-restart, auto-replication, and auto-scaling
  • Complete control of application life cycles, load balancing, and maintaining high availability

As mentioned above,  Kubernetes integrates with a myriad of third-party solutions, storage vendors, and orchestration tools. By integrating other open-source projects Kubernetes can be further enhanced to provide even greater benefits.

 

Kubernetes on Ridge

Ridge’s Kubernetes Service (RKS) is a fully certified, managed Kubernetes service. It is similar to Google’s GKE or Amazon’s EKS, but with added value via the Ridge Cloud global infrastructure. Ridge cloud is a collection of local data centers from around the world. As a result, developers have access to a global network of 100+ locations to deploy their clusters. Considerations can be based on location, infrastructure, and compliance.

Much like vanilla Kubernetes, RKS allows developers to focus on application development and deployment. The Ridge service offers an advantage on top: automatic cluster configuration. RKS is designed with usability and flexibility in mind. As a result, once a developer specifies cluster parameters, the Ridge Kubernetes Service will deal with configuration, installation, and maintenance, including any Kubernetes software and security updates. Since Ridge Kubernetes Service offers automatic cluster scaling, developer involvement is reduced to a minimum, and the system largely monitors and corrects itself.

The Ridge Kubernetes Service system is a powerful tool for edge computing. Latency-challenged locations often suffer from poor application performance and lack the necessary resources to develop an edge computing network. With Ridge Cloud, businesses can leverage Ridge’s network of world-class cloud big data centers closest to their users, and in the event of an outage, the Ridge Cloud will automatically adjust the workloads to ensure uninterrupted performance.

Organizations will also ensure they continue to meet strict compliance guidelines with Ridge Cloud. In the event of an outage or compliance violation, the Ridge Cloud will automatically adjust the workloads to a suitable data center. All Ridge’s data centers are conformity compliant with standards like SOX, ISO, and HIPAA. When developers set parameters, your containerized workloads running on RKS will be shifted to data centers that match your criteria.

Terminology—the basics

Cluster

Cluster: A set of nodes running containerized applications.

Container: A container is an instance of software running on physical or virtual infrastructure.

Pod: A Kubernetes Pod is a group of containers and is the smallest unit a Kubernetes deployment can administer. All containers in a Pod share the same memory and storage resources.

Nodes: A Kubernetes Node manages and runs multiple Pods in a group.

Control Plane

Control plane: A This is a collection of processes responsible for controlling the Kubernetes nodes. Any task assignments will begin here. Nodes will perform tasks assigned from here.

 

API server: The gateway to the Kubernetes cluster. The API Server implements a RESTful API, performs all API operations, and communicates with the defined endpoints.

 

Scheduler: The Scheduler is responsible for watching over the resource capacity and assigning work.

 

Ingress: An API object that manages external access to the services in a cluster, typically HTTP.

Ingress may provide load balancing, SSL termination and name-based virtual hosting.

 

Controller Manager: The Controller Manager is responsible for ensuring the cluster operates as expected and will respond to events accordingly.

 

etcd: Consistent and highly-available key value store used as Kubernetes’ backing store for all cluster data.

 

Kube scheduler: A Control plane component that watches for newly created pods with no assigned node and selects a node for them to run on.

kubectl: The kubectl command line tool lets you control Kubernetes clusters.

Frequently Asked Questions

Kubernetes vs Docker, what’s the difference?

These technologies actually complement one another, as Docker is focused on container orchestration and running on a single Node. Kubernetes deals with larger cluster orchestration, and as such the two technologies tend to work hand-in-hand.

What is container orchestration?

Container orchestration is the automatic process of managing or scheduling the work of individual containers for applications based on microservices within multiple clusters.

Who uses Kubernetes?

Nearly 80% of businesses use Kubernetes in production environments in nearly every industry around the world for easier application deployment and management of distributed services.

Is Kubernetes free?

Yes, Kubernetes is open-source software that can be used for free and is constantly updated and shared by thousands of developers across the world. It can be used by anyone without purchasing licenses.

Why is Kubernetes called k8s?

The 8 comes from replacing the “ubernete” of Kubernetes with an 8, hence k8s. This is a common spelling used to shorten Kubernetes for easier writing or mentioning when discussing containers.

 

Is Kubernetes a PaaS?

No. Kubernetes is a set of primitives that lends itself well to building PaaS tools, but it is not a PaaS. 

Can I create a cluster that spans multiple locations?

No. A cluster will have all its components in a single data center. You can create multiple clusters in multiple geographies, but with High Availability, you can reduce downtime if a cluster fails.

What is High Availability?

A High Availability cluster, or fail-over cluster, is a group of containers that can be utilized with minimal downtime. In the event of a disaster or crash, the new group of containers will provide continued service until the original cluster is restored.

What’s next:

What are Containers

Kubernetes vs. Docker – What’s the difference?

Learn more about Ridge Kubernetes Service

Get started with Ridge Cloud

Take a look at Ridge’s Kubernetes API


Author:
Linkedin Mati Lerner, Co-Founder & CEO | Ridge
Avatar photo
Mati is not only one of the founders of Ridge, but also a cloud expert. He enjoys writing about the benefits of cloud nativity & different cloud architectures.