About a year ago, I wrote a story on the potential battle between containers and serverless for control over the future of cloud computing.
While it was just a year ago, it was still at a time when containers were not the bigger threat to virtual machines (VMs) that they are today. Kubernetes was not the conversation dominator that it is today. And serverless might as well have been sending people to Mars.
But since then, the rise of Kubernetes has made containers a “have to have” at least in terms of future cloud plans, and serverless has quickly taken on the role of promising up-start. That last point has become very obvious as I can now ask people about serverless and not always get quizzical looks.
‘And,’ Not ‘either/or’
Instead of my original story that posted the “either/or” question when it came to containers and serverless, today’s reality has turned to focusing on the “and” aspect of containers and serverless. (And I should note that this “either/or” title is sort of stolen from a tweet by Enterprise Strategy Group Senior Analyst Edwin Yuen.)
Much of this can be traced to Kubernetes. While initially begun as a way to orchestrate container deployments, the Kubernetes framework has taken on a much broader appeal and bolstered confidence in containers. Perhaps the most prominent being AT&T using containers as the basis for the next iteration of its cloud infrastructure.
Serverless has also turned the corner. While actual production usage remains somewhat varied, it does exist. And that traction is gaining attention from the container community.
Google, at its recent Next event, unveiled the Knative initiative to unify the current disparate serverless deployment ecosystem. It does this by providing an open source set of components that allows for the building and deployment of container-based serverless applications that can be transported between cloud providers.
And the “K” part of “Knative” highlights that all of this will happen under the guise of Kubernetes. Basically, Kubernetes will orchestrate the running of serverless functions inside of a container.
Sure, there has been some backlash to the Knative approach – and the glaring omission at this point of work with Amazon Web Services (AWS) or Microsoft Azure – but it highlights the growing connection between containers and serverless. And it must also be noted that Pivotal, IBM, Red Hat, and SAP were all part of the initial Knative work.
More importantly, this effort is a strong step toward breaking down the vendor lock-in challenge that the serverless space is currently facing. Kubernetes, which evolved out of Google, came along and attacked this for the container space, and if nothing else Knative could be a good push to tackle this for serverless.
I understand that both containers and serverless remain a small percentage of overall cloud usage by enterprises, but the development aspect of both is rapidly advancing. And I am probably not going too far out on a limb in predicting that the adoption and integration of both will have progressed significantly over the next 12 months.
The more surprising aspect of that adoption could be that instead of enterprises choosing one or the other, they will be adopted as a package deal.