Although Kubernetes is still a relatively young technology, its adoption rate continues to rise. This study looks at the challenges teams face when using Kubernetes and the factors that influence technical decision‑makers.
Kubernetes just passed its fourth birthday, making it a very young technology, but adoption has been rapid. Many previous studies have used numbers and charts to show trends in cloud‑native tooling, but few have drilled into which factors influence decision‑makers and what difficulties teams encounter when using these tools.
Heptio therefore commissioned a third‑party firm to survey nearly 400 IT decision‑makers from a wide range of industries, in order to understand which teams/organizations are moving to Kubernetes, what obstacles they face, and how they are benefiting from the technology.
The 378 respondents to this survey are responsible for cloud‑native technology deployment and strategy. 37% work in organizations with 100–1000 software developers, and 23% in organizations with more than 1,000 developers.
Kubernetes adoption
The first question asked how respondents are using Kubernetes. 60% are using Kubernetes for container orchestration today. That figure is expected to grow to 65% next year, while usage of other orchestration platforms (Docker Swarm, Mesos, etc.) is expected to decline or remain flat.
There is a common misconception that containers are only useful for stateless applications. The survey shows that Kubernetes is now being used for a wide variety of workloads, including data‑intensive analytics, machine learning, and IoT applications.
Kubernetes has also increased overall confidence in open source. Compared with proprietary software, respondents prefer open‑source platforms, and they tend to favor open‑source solutions for logging and monitoring as well. For load balancing and storage, however, many still prefer commercial/closed‑source offerings.
30% of respondents are using Kubernetes in production. Among those using Kubernetes at all, roughly half have it running in production, and the larger the company, the more likely this is to be the case.
Across all teams running containers in production, 68% use Kubernetes. That is roughly twice the number using Docker Swarm. By next year, Kubernetes is expected to be in production for the majority of container workloads.
Multi‑cloud is no longer just a concept—it is already a common practice. Two‑thirds of respondents have adopted a multi‑cloud strategy. Reasons include: access to better features, more bargaining power with providers, lower operating costs, and improved performance.
Kubernetes is naturally well‑suited to multi‑cloud environments. By using the right management tools to provide a common operating model, Kubernetes enables the same application to be deployed on public clouds or on‑prem infrastructure. Public cloud providers, however, have little incentive to promote multi‑cloud, so the key to success in multi‑cloud deployments is avoiding lock‑in to a single vendor.
Kubernetes is still young, and its ecosystem is still maturing. Features and surrounding tools are evolving at different speeds, which means teams face various challenges when evaluating and running Kubernetes. Major pain points include: early design and deployment decisions (47%), aligning internal teams on a common technical understanding (47%), lack of in‑house experts (45%), ensuring reliability for mission‑critical workloads (43%), and managing complex policies (42%).
Looking at the CNCF landscape, with its huge range of projects, it’s easy to see why making architectural decisions around Kubernetes is hard: there are simply too many projects and tools, and teams need guidance.
When asked what help they most need with Kubernetes, 41% of respondents answered “architecture design”. Architecture is indeed a pain point for many teams, as it involves many considerations: evaluating physical and software infrastructure, integrating new technologies, and adopting new ways of operating. The second‑most common answer was “deployment” (16%), and 14% said they need help with day‑to‑day Kubernetes operations.
Teams rely on a variety of sources for guidance, including orchestration vendors (55%), independent consultancies (42%), industry peers (54%), and community forums (53%). Only 4% of respondents said they needed no help at all when deploying Kubernetes. This suggests that many organizations still struggle to find or access the support they need.
Who actually drives Kubernetes architecture decisions? Responses varied: architecture teams, IT operations, SRE, cross‑functional teams, business units, and external consultants all appear. In many organizations, it is still unclear who ultimately owns Kubernetes. Unsurprisingly, internal alignment around common technical understanding remains a major challenge.
The survey also shows that larger organizations are more likely to adopt Site Reliability Engineering (SRE) practices. SRE combines aspects of IT operations and software engineering to ensure reliability and scalability.
The results indicate that 31% of teams use SRE‑Ops to make decisions, 38% of teams running Kubernetes in production use SRE, and among teams with more than 1,000 developers, 48% use SRE practices.
We believe SRE‑Ops plays a critical role in environments where DevOps needs to scale.
What capabilities do teams most need when using Kubernetes?
Most respondents want better visibility into what is happening across their Kubernetes footprint.
Specifically, teams with more than 1,000 developers are more likely to want real‑time performance visibility and better access to historical logs, audit records, and inspection data.
Regardless of where a team is on its Kubernetes adoption journey, it is worth reviewing the solutions available in the Kubernetes ecosystem and considering which additional capabilities can help them run Kubernetes more effectively.
The most‑cited needs include: visibility (58%), simplified cluster support and recovery (55%), continuous consistency testing (49%), integrated support and troubleshooting (49%), and access to historical logs, audit, and inspection data (45%).
If your team is currently evaluating Kubernetes, there are some useful lessons from those already running it in production.
Among teams using Kubernetes in production, 65% want better real‑time visibility into cluster performance. For teams running more than 1,000 containers, that figure rises to 69%. As container counts increase, more clusters are needed, and the demand for visibility grows accordingly.
The ability to access audit logs and historical inspection data is also critical. 65% of respondents running Kubernetes in production consider this a key capability. Many also highlight the importance of standardization or consistency testing and simple, reliable backup.
Is it worth overcoming all these challenges to deploy Kubernetes?
The survey suggests the answer is clearly “yes”. Teams running Kubernetes in production are more likely to report three key benefits: shorter software delivery cycles, improved resource utilization, and greater use of complementary open‑source technologies.
For many organizations, the primary goal is to shorten development cycles while improving resource efficiency. Kubernetes and its surrounding ecosystem can help achieve these goals.
The main benefits cited for using Kubernetes are: the ability to use complementary technologies, faster software delivery cycles, and better resource utilization (with red indicating the average and yellow representing teams running Kubernetes in production in the original charts).
Kubernetes is increasingly being used in production, and its ecosystem is evolving rapidly, with new tools and solutions appearing all the time. While users still face a number of challenges, in the long run the potential benefits are likely to far outweigh the difficulties.
To further advance Kubernetes adoption, the key requirement is better visibility. This underscores the importance of upstream Kubernetes as an independent, cloud‑agnostic platform that helps avoid vendor lock‑in. Such a platform allows teams to take full advantage of best‑of‑breed open‑source technologies while retaining maximum flexibility in a fast‑changing ecosystem.
If you are considering Kubernetes or are in the evaluation/deployment phase, it is worth seeking advice from teams that are already running it in production. Cloud‑native technologies (including Kubernetes) are challenging, but for teams that prioritize software delivery, want to shorten development cycles, improve resource efficiency, and access the best open‑source technologies, the effort is clearly worthwhile.
Technical support
Scan the code to join the technical support WeChat group
Official account
Scan the code to follow and get the latest updates