The State of Enterprise Kubernetes

At Kasten by Veeam, we were curious to learn how it’s been tuned to better meet the needs of developers. This past year, DZone developed the Enterprise Kubernetes Report, which includes survey data about how teams are deploying and using Kubernetes, and their challenges in implementing real-world clusters at scale. The report includes insights for building a solid Kubernetes strategy, complete with security and observability for simplifying maintenance and monitoring.

In this post, I’ll summarize some key takeaways from the report around:

• The state of Kubernetes adoption
• How Kubernetes is being leveraged in the enterprise
• Cloud-native observability
• Benefits of CI/CD pipelines in the Kubernetes ecosystem

The State of Kubernetes Adoption

In the past, developers used big monolithic applications that were complex to manage and use. Pushing changes into production could take months, and progress was often interrupted with downtime. Microservices changed the game, breaking down those applications into multiple, individual services. First Docker was introduced as a platform for deploying these services via containers; then Kubernetes overtook Docker. 

Kubernetes makes life easier for developers by helping them manage workload more efficiently. It eliminates downtime, increases scalability and security, allows for roll backs if something goes wrong, and increases productivity via automation. With all of these benefits, it’s no wonder its popularity has soared.

To compile this report, DZone surveyed software developers, architects and other IT professionals, and received nearly 470 responses. Here are some key stats the data revealed:

• Container adoption remains at or around the 90% mark
• Kubernetes adoption has stalled around 73%
• 14.8% of Kubernetes deployments are in development, and 17% are in production
• 85.5% of participants from larger companies said they ran Kubernetes clusters, whereas 63.4% of respondents from smaller companies said they did
• More than half of respondents (52%) are running stateful workloads in Kubernetes

What’s also interesting is the array of applications that developers run on Kubernetes. Take a look at the graph below:

Kubernetes is ideal for everything from web apps to CPU-, GPI- and memory-intensive applications, to storage and general compute functions. And scalability is key – Kubernetes can handle highly variable traffic without overcommitting resources:

Kubernetes in the Enterprise

With these advantages, it’s no wonder enterprises are jumping on the Kubernetes bandwagon. DZone defines the enterprise as For them, Kubernetes is recognized as the “server of the serverless,” enabling support for compliance and regulations, data sovereignty and granular control. 

There are several other reasons why Kubernetes is gaining traction in the enterprise. The platform offers a rich feature set, out of the box, so developers can quickly set up and manage containers. Because it’s open source, development is driven by community members whose extensions, tools and services enhance Kubernetes and make it more powerful. Kubernetes is also compatible with most standard programming languages and operating systems, so adoption doesn’t require an overhaul. 

Enterprises can leverage Kubernetes for: 

• Serverless computing: Knative (kn), recently introduced by the CNCF, abstracts Kubernetes as Serverless computing, eliminating repetitive configuration and build tasks, and threatening to disrupt on-premises serverless options for building event-based architecture.
• Big data processing and stateful data services in data science and machine learning (ML) tech stacks. Containers on Kubernetes enable the flexible deployment of big data software without the complexity and high overhead of traditional data cluster setups. Instead, Kubernetes leverages on-demand GPU/CPU compute resources for ML processing.
• Sophisticated container-based cluster federation.Using Kubernetes, organizations can combine a hybrid or multi-cloud setup into a single platform. This enables them to avoid vendor lock-in by providing flexibility around hybrid and multi-cloud clusters.
• DevOps enablement:Kubernetes can be used alongside open source tools such as Helm IaC, Grafana telemetry dashboard and alerting, and Prometheus metric ingester to assist with operations.

What about Cloud Native Observability?

One of the contributors to the report, Marija Naumovska, Co-Founder and Technical Writer at Microtica, offered her perspective on observability in cloud-native environments, such as Kubernetes. Observability helps developers understand what’s happening in the clusters, so they can optimize compute power and performance. Observability also helps manage costs and keep clusters secure.

However, there are several challenges with achieving observability in Kubernetes. It has more components than traditional infrastructures, each of which produces its own metrics. Modern applications can include many microservices, as well, and these may be distributed across multiple clusters and instances. What’s more, containers are dynamic and their lifecycles vary. Advances in observability, such as  real-time monitoring, performance analytics, and application visibility, are helping to solve these challenges.

In the report, Naumovska provides an overview of popular observability tools – Prometheus, Open Telemetry, Jaeger and cAdvisor – including their relative advantages and disadvantages.

Helm Charts & Best Practices

In addition to in-depth discussions about the state of Kubernetes adoption, Kubernetes in the enterprise, the need for (and challenges around) observability and other key topics, the report includes an Advanced Guide to Helm Charts for Package Management in Kubernetes, written by Yitaek Hwang, Software Engineer at NYDIG. Now the de facto package manager for Kubernetes, Helm adoption is growing in sync with the orchestration platform, and is used by about two thirds of respondents today. (Check out our new lab on Helm at KubeCampus.)

In the guide, Hwang reviews Helm structure and lifecycle, as well as key best practices, such as how to bootstrap charts, check for errors, package with subcharts and use annotations. He also explores the limitations of Helm with the caveat that it’s easier than ever to modify and install various Kubernetes components with Helm, thanks to a wide selection of public, production-ready Helm charts.

Benefits of CI/CD Pipelines in a Kubernetes Ecosystem

Finally, Sudip Sengupta, Technical Writer at Javelynn, contributed a section on CI/CD approaches in a Kubernetes ecosystem, including some best practices for implementing an efficient CI/CD framework. According to Sengupta, “Kubernetes provides a combination of deployment objects, rollout strategies, and resource controllers to help automate the build, test, and integration processes of a CI/CD pipeline.” It also provides algorithms that help with scheduling batch jobs, enabling gradual rollouts and autoscaling workloads.

Kubernetes supports pull-based and push-based pipelines for implementing configuration changes in a workflow:

• Pull-base – code and infrastructure changes are applied within the cluster
• Push-based – configuration changes are committed in a repository that adds commits to an executed pipeline

While pull-based pipelines ensure continuous security and make it easier to scale identical deployments, these only work for Kubernetes deployments and each Kubernetes cluster has to adopt this framework. Push-based pipelines are simpler to implement and offer more flexibility, and they’re ideal for hybrid/multi-cluster environments. However, they require external administrative access to the CI/CD pipeline which may compromise security.

Regardless of which CI/CD pipeline framework you use, using CI/CD pipelines in a Kubernetes ecosystem provides several benefits, including shorter release cycles, the ability to autoscale and replicate workloads, efficient node usage and microservices-based deployments that provide for simpler, application deployment and management. Check out these best practices before getting started. 

Additional In-depth Discussions

While those are some of the highlights of the report, you’ll also find in-depth discussion on other topics such as: 

• The Importance of Kubernetes in the Cloud Native Ecosystem
• Architectural Patterns for Microservices With Kubernetes
• An Assessment of Kubernetes and Machine Learning
• Kubernetes Cost Optimization

You’ll also find a summary of how automotive software vendor, Zenzeact, accelerated innovation in the area of self-driving by streamlining Kubernetes backup management with Kasten K10, Kasten by Veeam’s solution for secure and reliable Kubernetes data and application backup and recovery. Zenseact has deployed Kasten K10 across roughly 60 nodes, with about 1.5 petabytes of NVME-based storage, and uses Red Hat OpenShift Container Storage along with Dell Isilon S3 to store the backups and a bare metal cluster for OpenShift. Using Kasten K10, the company is able to perform automatic backups and save about 150 hours per month on backup management.

Finally, at the end of the report, you’ll find an array of recommended resources for diving deeper into the topic of Kubernetes, and an extensive solutions directory listing various solutions for improving the developer experience with the Kubernetes platform. Of course Kasten K10 by Veeam is included in the directory, representing the industry’s leading backup and recovery tool for Kubernetes data and applications. Kasten K10 is purpose-built for Kubernetes, providing cloud native data management for Day 2 operations. Enterprises benefit from an easy-to-use, scalable and secure system for Kubernetes application backup/restore, DR and mobility. 

Download the Report!

Now that you’ve read the highlights, download the Enterprise Kubernetes Report to learn all the metrics and details we uncovered in our research. If you’re interested in testing our enterprise-grade Kubernetes backup and recovery solution, Kasten K10, just click the link  below: 

Kubernetes Enterprise Trial