What is Kubernetes - Container Orchestration Software Explained

by BoxBoat | Saturday, Jan 26, 2019 | Docker Kubernetes

Developed by Google in 2014, Kubernetes (or K8S) is an open source platform for managing containerized workloads and services on public, private, or hybrid clouds.

Kubernetes is a very popular container orchestration tool and has a large and rapidly growing ecosystem. There are many services and tools available that support Kubernetes.

But what exactly is Kubernetes? Why do you need it? And what does it do?

Why Do You Need Kubernetes?

When deploying applications as containerized images, there are many manual processes involved when it comes to management and scheduling of these containerized services at scale. For enterprise containerized applications, there are hundreds if not more in highly available configurations, of containers that make up these services to continually manage. Manually scaling and deploying each of these services in an enterprise environment would take a lot of time!

Kubernetes eliminates these manual processes by giving you the orchestration and management capabilities to run and deploy your containers at scale.

A Kubernetes orchestration allows you to develop application services that span multiple containers, manage those containers, and even schedule them across a cluster of machines.

Of course, Kubernetes container orchestration doesn’t come without its own cost of overhead and complexity. Development shops with smaller projects should look to alternative platforms like Docker Swarm, or even stand-alone Docker if high availability is not a paramount requirement.

However, as soon as you begin to scale this application to a production environment or begin to deploy multiple applications, you will end up with a high volume of containers that need to work together to deliver various individual microservices, adding additional layers of complexity in the process.

Kubernetes mitigates the majority of the problems associated with container proliferation by sorting these containers into “pods”. Pods provide a layer of abstraction to grouped containers, helping you to schedule workloads and allocate any necessary resources, such as networking and storage, in an effective manner.

What Can You Do With Kubernetes?

As mentioned, Kubernetes helps to orchestrate and automate your containers so that it can run efficiently. But to get a clear understanding of what Kubernetes can do, here’s a breakdown:

• Orchestrate your containers across multiple hosts
• Ensure that the “desired state” is maintained with total number of containers available, or autoscaled appropriately via Horizontal Pod Autoscaling (HPA)
• Make effective use of your hardware resources to run your enterprise applications
• Manage and automate container deployments and updates
• Scale containerized applications on the go - you can do this on a local machine
• Mount and assign more storage to operate stateful apps
• Run regular health checks of your containers with autoplacement, autoreplication, autorestart, and autoscaling

However, to get the full advantage of Kubernetes, and to do most of the tasks mentioned above, Kubernetes will need to work with other open source projects, which includes:

• CI/CD (we like Jenkins for its flexibility)
• Docker Image Registry (Harbor, DTR, Artifactory)
• Telemetry (prometheus, heapster, kibana, elastic)
• Software Defined Networking (Calico, WeaveNet)
• Security & RBAC (Vault, Clair, Aqua, Twistlock, LDAP integration)
• Automation (Ansible, Chef, etc.)

As you can see, there are many components involved to an end-to-end container orchestrated solution. Having a tool like BoxOps will help you manage your DevOps tools in one place, accelerating your deployments to Kubernetes, and preventing your team from having to use multiple logins, which can hinder productivity.

Understanding the Terms Associated with Kubernetes

Similar to any popular technology solution, there is technical jargon that can become a barrier to entry. Here, we share some common terms to help you better understand Kubernetes:

Master

The machine interface that controls the Kubernetes nodes and assigns tasks to them. There are a numnber of components that run under the hood that make up the master nodes.

Node

A node is a worker machine that performs the requested tasks. The Master nodes controls these worker nodes via additional components that run on the workers.

Pod

The smallest atomic unit in Kubernetes. A pod often houses a single container, but can house multiple containers that always reside together. All the containers in the pod share the same hostname, IP address, IPC, and other resources. The pods abstract the storage and network away from the underlying container, which allows them to become portable so you can move them around the cluster easily.

Replication controller

Controls how many concurrent copies of a pod that should be running on the cluster. There are a number of controller types including replicasets and daemonsets.

Kubernetes Service Proxy

Kubernetes service proxies automatically deliver service requests to the right pod, no matter where it has been relocated (or replaced) in the cluster. This goes for both the management plane and workloads.

Kubelet

A service that runs on nodes and reads the container manifest. It ensures the defined containers have been initiated and are running. The kubelet runs on every node and performs a control loop to ensure that the desired state is kept.

Kubectl

The primary end-user command line interface for Kubernetes.

Where Does Kubernetes Fit in the Container Infrastructure

Source: Redhat.com

Kubernetes sits on top of the host operating system where it interacts with the pods running on the nodes. The master receives instructions from an administrator, or DevOps teams, which is then relayed to the nodes.

During this delegation, Kubernetes works with other services to decide which node will be best for the assigned task. Then, once it has decided which node to go for, Kubernetes then allocates the resources to the node to fulfill the requested work.

From a high level, Kubernetes provides better control over your services without having to micromanage each service container separately. It is mainly a case of assigning a Kubernetes master and defining both the nodes and the pods.

Where Does Docker Come In?

Kubernetes does not replace Docker. In fact, the two work together.

Whenever Kubernetes schedules a pod to a node, the kubelet on that node will request Docker (containerd) to instantiate those containers within the pod. The kubelet then proceeds to continuously receive status updates from those containers that Docker is running and then amalgamates all that information in the Kubernetes master. Docker pulls the containers onto the node and runs them as normal.

The difference that we see here is that an automated system (Kubernetes) is asking Docker to run specific nodes to execute a particular function or application instead of an admin having to run these nodes through manual processes.

Kubernetes is Just One Piece of a Containerization Puzzle

Containerized services have a multitude of benefits from faster load time to portability, but the problem arises when you're running massive enterprise applications in this fashion - you will end up having to look after thousands of containers.

Kubernetes plays a pivotal role in helping you manage a large number of services and respective containers, saving your development team from a lot of manual labor. But to comprehensively implement a container orchestration, Kubernetes has to work with other tools as well, which can lead to complications and poor productivity.

Having a container solution management platform, like BoxOps, can centralize all your tools on a single interface, and your team only needs to log in once.

For more information on Kubernetes or other container technologies, please get in touch with a BoxBoat DevOps specialist today.