What Do Customers Want From The Kubernetes Ecosystem In 2020


During the last two years, Kubernetes has matured to become the foundation of modern infrastructure. It’s not only the fastest-growing open source project but also becoming the most widely adopted infrastructure technology in enterprises. Major hybrid cloud platforms are now powered by Kubernetes.

SEE: How Kubernetes Has Changed The Face Of Hybrid Cloud

While the adoption is growing at a rapid pace, there are two key challenges with Kubernetes – Lack of clear demarcation between the developer and operator experience and lack of application context.

Dev vs Ops

Today In: Innovation
The architects of Kubernetes tackled some of the hardest problems associated with scalable infrastructure. Based on containers, Kubernetes delivers an unmatched scale and reliability of workloads. Fundamentally, Kubernetes is designed for system administrators and operators that deal with compute, network and storage infrastructure. An enterprise IT team may provision a production-grade Kubernetes cluster by assembling existing building blocks available within the data center. But even after making the Kubernetes cluster available to internal developers, the operations team will have to work closely in deploying and running the applications. Typically, in traditional IaaS and PaaS environments, developers own the application lifecycle management.

Kubernetes blurs the fine line that exists between developers and operators. While this may seem to be an advantage, it has many drawbacks. Irrespective of the persona – developer or operator – every user of Kubernetes needs to have the same level of understanding and expertise in dealing with the cluster. From packaging the workloads to configuring resources to scaling the deployments, developers and operators have equal involvement.


| Paid Program
5 Ways To Secure Your Business In A Multi-Cloud World
| Paid Program
The Art Of A Smart City
| Paid Program
Investing In Data Culture To Harness The Opportunities Presented By Asia Pacific’s Digital Economy
Developers prefer to focus on the rapid iteration of code delivering new features and fixing known issues. Operators are measured on the overall stability of the infrastructure and applications. This model encourages the clean separation of the personas with clearly defined roles and responsibilities. With Kubernetes, it is hard to define where the developer role ends and where the operator role starts.

Going forward, Kubernetes customers want an enhanced developer experience that abstracts the nuts and bolts of the infrastructure while exposing the most essential capabilities expected by developers.

Application Context

When an application is deployed in Kubernetes, it’s broken down into a collection of objects. For example, a simple web application running in Kubernetes has a Deployment object to run the frontend, a StatefulSet object for a highly available database instance, ConfigMap to define the configuration settings of the database server, Secrets to hold the passwords used with the database, Persistent Volumes and Claims for storage, a Service to expose the database to the web frontend and an Ingress to make the web app available to the outside world.

A barebones workload deployed in a cluster typically translates to a dozen objects mapped to various Kubernetes primitives. Each object is independent of other objects. Kubernetes has no knowledge of objects that belong to the same application which may need to be treated as a single logical unit.

Kubernetes lacks the application context which makes it hard to manage the lifecycle efficiently. Even though package managers such as Helm attempt to bring context to deployments, that’s mostly confined to the declaration of the application. Once an application is deployed in a cluster either through Helm or through the command line tools, it loses the context.

As Kubernetes makes its way to the enterprise data center to run the line of business applications, application context and awareness become the key. Customers want a mechanism to declare, package, and manage a unified application entity rather than dealing with loosely coupled resources running in a cluster.

Notify of

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Inline Feedbacks
View all comments
Would love your thoughts, please comment.x