There’s general agreement that containers have become one of the fastest emerging technologies to be adopted in the enterprise in recent memory. The debate now concerns that rate at which containers adoption is continuing to grow. A new survey of 540 enterprise developers published by the Cloud Foundry Foundation (CFF) finds that 25 percent of organizations are employing containers. While that’s impressive, that number only represents a 3 percent gain over the previous year. However, the number of organizations evaluating containers in 2017 is 42 percent, an 11 percent gain over the previous year.
No matter how anyone assesses container adoption, Devin Davis, head of marketing and research for CFF, says the fact that 25 percent of organizations are already using containers within a comparative short period of time is a major achievement when compared to almost any other technology implemented in the enterprise.
In terms of managing those containers, 53 percent of companies were managing them either via a container service, such as Amazon Web Services (AWS) EC2 or Google Container Engine, or a provider-managed platform-as-a-service (PaaS) environment such as IBM Bluemix or Microsoft Azure. Only 15 percent were using self-managed container orchestration tools. Of that group, 27 percent were using CoreOS Tectonic, a commercial distribution of Kubernetes, followed by Hashicorp Nomad, Docker Swarm and Mesos, each around 20 percent.
In general, Davis says it’s clear that momentum behind Kubernetes is accelerating. Because of that, the CFF has begun to invest in projects focused on Kubernetes as well as the Cloud Foundry PaaS. The CFF is now driving the development of Kubo, an open-source project from automating the provisioning of Kubernetes clusters based on BOSH software originally developed for Cloud Foundry PaaS environments.
IT organizations are being asked to consume more emerging backend technologies at a faster rate than ever before. For, example, new serverless frameworks based on a function programming model will be deployed alongside containers. In some cases, stateless applications that were the province of first-generation container applications are good candidates to deploy on an event-driven serverless framework. At the same time, containers are maturing to the point where they can be used to deploy a stateful application in a production environment. At some point in the future, Davis notes, it’s probable that serverless frameworks will be deployed on Kubernetes to provide a common framework. In fact, containers may begin invoking functions running on a serverless framework to run any number of child processes when required.
Unfortunately, history indicates when confronted with too many choices at once, enterprise IT organizations tend to freeze for fear of making a wrong choice. In truth, there may be no such thing as a wrong choice. IT organizations are going to wind up employing a mix of backend architectures to run different classes of application workloads. The next big challenge will be making it possible to employ machine and deep learning algorithms to enable workloads to self-identify where they should run based on the characteristics and attributes of the application.