Why Kubernetes Is the Future OS: Expert Insights, Learning Path, and Real‑World Troubleshooting

1 What is Kubernetes?

Kubernetes is presented as the future operating system for cloud‑based IT infrastructure. Companies will allocate cloud resources into Kubernetes clusters for different business units, and service mesh will become a foundational layer alongside the cluster.

Alibaba has migrated roughly half of its workloads to custom Kubernetes clusters and aims to run 100% of its services on Kubernetes this year. Service mesh is already in production in some Alibaba divisions such as Ant Financial.

In the coming years, Kubernetes will be as ubiquitous as Linux.

2 Kubernetes vs. Traditional Operating Systems

Traditional OSes abstract hardware resources like CPU and memory. Kubernetes abstracts a pool of machines (each with its own OS) and provides a unified API for containerized applications, which bundle their dependencies and run independent of the host OS.

3 Kubernetes and Google’s Borg (Site Reliability Engineering)

Kubernetes inherits core ideas from Google’s Borg system, which the book "Site Reliability Engineering" describes. Understanding both the operational methods and the underlying system (Borg/Kubernetes) is essential.

4 How to Learn Kubernetes?

Kubernetes has a steep learning curve because it spans kernels, virtualization, containers, SDN, storage, security, and trusted computing. It also integrates many cloud services (ECS, VPC, Load Balancer, Security Groups, Log Service, Cloud Monitor, Service Mesh, etc.).

Effective learning follows three steps: understand the technology evolution and landscape, get hands‑on experience, and continuously reflect on the underlying principles.

5 Controller Analogy – The Refrigerator

Controllers in Kubernetes implement a declarative, desired‑state model. The article likens this to a refrigerator that maintains a set temperature and turns on lights when opened—users request a state, and the controller makes it happen.

6 Why a Namespace May Not Delete

When a namespace is marked for deletion, the controller must clean up all resources across every API group. If any resource remains, the namespace stays in "Terminating" state, often due to network or permission issues between the API server and extensions.

In Alibaba Cloud, VPC routing must include the container network segment; missing routes can block the cleanup process. The routing controller adds these routes when new nodes join, using RAM role permissions.

At the end of the article, a free e‑book "Deep Dive into Kubernetes" is offered for download.