This article is based on reporting originally published on Forbes.com (2020). Statistics and survey references originate from the source article and have not been independently verified.
Kubernetes Beyond the Basics
About five years after Google handed over control of Kubernetes to the Cloud Native Computing Foundation (CNCF), the platform had already become the de facto standard for container orchestration. A 2019 survey conducted in partnership with Aqua Security reportedly found that 87% of respondents were running applications in containers, with nearly 90% of those running containers in production.
What’s less commonly discussed is how Kubernetes is being used in ways that go well beyond its original purpose of scheduling and orchestrating containerized workloads. Developers, vendors, and enterprise users are actively extending Kubernetes to serve other critical functions across the stack. Here are five of the most significant—and perhaps surprising—enterprise use cases.
1. On-Premises Deployments, Not Just Cloud
Kubernetes is frequently associated with public cloud providers, but many enterprises cannot—or choose not to—move sensitive data or proprietary intellectual property to a public cloud. Others have significant on-premises infrastructure they want to continue using.
Kubernetes handles both scenarios equally well. IT teams can use the same tooling, the same deployment pipelines, and the same operational practices whether they are running workloads on AWS, Azure, GCP, or in a private data center. Cloud-native application architectures—built around containers and microservices—provide the same benefits of speed and scalability in private cloud environments as they do in the public cloud.
This flexibility is a key reason Kubernetes adoption has spread across regulated industries such as financial services, healthcare, and government, where data residency and control are non-negotiable requirements.
2. Running Stateful Applications Alongside Stateless Ones
Early container use cases focused almost exclusively on stateless applications—services where each request is independent and no session data persists between calls. This limitation made containers impractical for many traditional enterprise workloads that depend on databases and other stateful systems.
The Kubernetes ecosystem has addressed this gap significantly. Features such as StatefulSets, PersistentVolumes, and PersistentVolumeClaims give operators fine-grained control over how storage is provisioned, attached, and retained for stateful workloads. Operators—custom Kubernetes controllers—have further extended this capability for specific databases and data platforms, automating tasks like provisioning, scaling, backup, and failover.
IDC data cited in Container Journal suggested that 76% of enterprises can now use containers broadly for mission-critical applications—a figure that would have been unthinkable when containers lacked meaningful stateful support.
3. Full Application Lifecycle Management
Kubernetes is increasingly used not just to deploy applications, but to manage them throughout their entire lifecycle—including high availability, disaster recovery, backup, and compliance.
Consider a traditional backup process in a database-centric environment: a restore operation typically recovers an entire machine image, bringing back all applications and data present at that snapshot point in time. Kubernetes-native backup tools, by contrast, can target specific application namespaces, workloads, or even point-in-time data states, making restores faster, more precise, and less resource-intensive.
Using a single platform for both application deployment and lifecycle management reduces operational complexity. Teams do not need to maintain separate toolchains for deploying applications and managing their long-term operational health.
4. A Unified Control Plane for Infrastructure
Kubernetes was originally designed to orchestrate containerized applications. Increasingly, it is also being used to manage the underlying infrastructure those applications depend on—storage, networking, and compute.
Two CNCF specifications make this possible:
- Container Storage Interface (CSI): A standardized API that allows storage vendors to expose their systems to Kubernetes without modifying the core Kubernetes codebase. This enables dynamic provisioning, snapshotting, and resizing of storage volumes across a wide range of backends.
- Container Network Interface (CNI): A standardized plugin model for configuring network interfaces in containers, allowing network vendors and open-source projects to integrate with Kubernetes in a consistent way.
Together, these interfaces allow Kubernetes to act as a unified control plane across both the application layer and the infrastructure layer. This reduces the number of separate management systems an organization needs to operate and keeps infrastructure management in step with the rapid pace of application-layer changes.
5. Accelerating AI and Machine Learning Workloads
Kubernetes is well suited for AI and machine learning pipelines for several reasons:
- Isolated environments: Data scientists can work in separate Kubernetes namespaces or pods, experimenting with different datasets or model configurations without interfering with each other or with production systems.
- Independent scaling: Kubernetes allows compute resources (including GPU-backed nodes) to scale independently from storage or other services, which is particularly valuable during the training phase of ML models, when GPU demand spikes significantly.
- Streamlined deployment: Once a model is ready for production, Kubernetes enables it to be packaged and deployed as a containerized microservice, using the same CI/CD pipelines used for other application workloads.
- Workflow automation: Projects such as Kubeflow—built on top of Kubernetes—automate ML pipeline steps including data preparation, training, hyperparameter tuning, and serving.
Major vendors including Google and HPE have built AI-specific offerings on top of Kubernetes, reflecting broader industry recognition that container orchestration is a natural fit for the iterative, compute-intensive nature of machine learning development.
What This Means for Enterprise IT
The common thread across all five use cases is convergence. Rather than operating separate platforms for application deployment, infrastructure management, data lifecycle management, and AI workloads, enterprises are increasingly using Kubernetes as a single, extensible platform that spans all of these concerns.
This consolidation offers real operational advantages: fewer tools to learn, fewer integration points to maintain, and a single community and ecosystem to draw from when solving new problems. The same speed, flexibility, and scale that made containers attractive for application development are now being extended into adjacent areas of enterprise IT—and Kubernetes is the platform making that possible.
As the open-source ecosystem around Kubernetes continues to mature, it is reasonable to expect further extension of the platform into areas such as edge computing, serverless workloads, and multi-cluster federation—each of which is already seeing active development in the CNCF project landscape.