A survey of 1,106 IT professionals published this week by Rancher Labs finds Kubernetes clusters deployed by IT organizations are starting to multiply.
The survey finds 91% of respondents are running more than one cluster across a mix of on-premises and IT environments, with 68% running somewhere between two and 10 clusters. Roughly two-thirds of respondents (65%) said they would also be interested in some form of ability to federate Kubernetes clusters. The primary reasons cited for running multiple clusters are workload/tenant isolation (70%), followed by team autonomy (48%).
Overall, a full 85% of respondents said they are running containers in production environments, while 91% said they are employing Kubernetes for container orchestration.
Peter Smails, chief marketing officer for Rancher Labs, says given that level of adoption, it’s clear the barriers to Kubernetes adoption are dropping. The challenge now is making it easier to adopt additional platforms such as Prometheus monitoring tools and service mesh platforms that the adoption of Kubernetes naturally begets as each Kubernetes environment continues to evolve, says Smails.
The survey also finds 71% of respondents are running on-premises workloads, while 67% are running cloud-based applications and 15% are running edge computing workloads. Two-thirds (66%) are deploying those containerized workloads on virtual machines, followed by public clouds (58%), bare-metal servers (44%) and private clouds (43%). About one-quarter (27%) said they were using any kind of platform-as-a-service (PaaS) environment, with Red Hat OpenShift the most widely employed (27%).
The dominant cloud services provider for running these workloads is Amazon Web Services (73%), followed by Microsoft Azure (41%) and Google Cloud Platform (39%).
The primary drivers of containers and Kubernetes adoption are IT architects, developers and DevOps teams (69%), followed distantly by IT operations teams (30%). The most common use cases for containers cited are internal applications (73%), designing microservices (71%), building customer-facing applications (70%) and modernizing legacy applications (53%).
A full 79% also note they have embraced continuous deployment within their Kubernetes environments, with Jenkins (51%) narrowly edging out GitLab (49%) as the dominant continuous integration/continuous delivery (CI/CD) platform. Not surprisingly, 92% said they are monitoring Kubernetes clusters, with open source Prometheus/Grafana being employed the most (60%).
As more organizations embrace multiple Kubernetes clusters, it’s apparent the challenges that come with managing Kubernetes clusters at scale are starting to manifest themselves. It may be comparatively simple to stand up a single Kubernetes cluster, but once those clusters start to multiply the need to implement best DevOps practices consistently becomes more apparent. In many cases, the adoption of Kubernetes will force organizations to accelerate the rate at which they are moving up the DevOps maturity curve.
In the meantime, as Kubernetes clusters become less exotic in IT environments, the tenor of the IT conversation is starting to shift. Instead of wondering just how much of a good thing Kubernetes might be, the issue now is whether Kubernetes is about to become too much of a good thing.