Why do we use Kubernetes?

Before the advent of Kubernetes, container management was predominantly a manual process, relying on human intervention for handling images and their replicas. This approach posed significant challenges, some of which are as follows:

• Handling situations where a system scaled down or experienced failures, and dealing with requests directed towards that system, required extensive manual effort and lacked automated solutions.

• Managing multiple replicas, ensuring their availability, and guaranteeing the smooth functioning of containers became increasingly complex when the number of resources and traffic kept increasing.

• Scaling down unnecessary instances and performing container restarts were unmanageable tasks managed manually, leading to inefficiencies and operational difficulties.

• Making changes to deployments and reflecting those changes in the Pods can become problematic without proper handling. Immediate restarts or reflections can disrupt the system’s stability and lead to unintended consequences.

Given these challenges, the emergence of Kubernetes addressed these critical needs and provided a comprehensive solution to streamline container management.

Addressing challenges with Kubernetes

Kubernetes works over the top of docker to address multiple problems that need to be handled manually through automation. Some of the solutions are stated below:

Automated handling of failures

With constant monitoring, Kubernetes detects failures or unresponsive containers and automatically initiates container restarts. When node failures occur, the control plane promptly addresses the situation by attempting multiple restarts of the containers or replacing them with healthy replicas, adhering to the specified replica count in the deployment configuration.

Automated rollbacks

Automated rollbacks in Kubernetes involve monitoring new replicas during the deployment update, detecting issues, and triggering an automatic rollback process by scaling up the previous stable replicas. This ensures proper deployment of changes across all instances while considering the liveliness and readiness of Pods, enabling a gradual rollback process if necessary.

Replication control

To manage multiple replicas, ensure their availability, and guarantee the smooth functioning of containers, Kubernetes has a replica controller (replica sets), which refers to creating and managing multiple replicas of a Pod or a set of Pods. It ensures that the availability of the desired number of identical Pod instances, as specified in the configuration, is running and available within the cluster. The purpose of replication is to enhance the application’s availability, fault tolerance, and scalability by distributing the workload across multiple instances.

In conclusion, Kubernetes is necessary to automate the process of containerisation and management.