Serverless: Is It The Kubernetes Killer?

Source:-forbes.com

A few years ago, many believed OpenStack was going to take over the world. It was inevitable, and no one questioned it. Docker shifted the landscape, and then along came Kubernetes, which was another game-changer. Today, everyone is excitedly talking about serverless (functions as a service). Did Kubernetes kill the OpenStack engineer? Is serverless going to kill Kubernetes?

The pace of innovation in the cloud over the last five years has been astounding to watch. There is no arguing that the technology is moving at a breakneck pace. The best practices of 2015 are so outdated it’s ridiculous. Any company that has been in business for even 20 years has replatformed at least a handful of times.

And in a few years, there will be compelling reasons to do so again. Some are arguing that companies should go all in and write the entire application stack with serverless. But there is no reason to continually throw the baby out with the bathwater. Just add new kinds of fancy bubbly water and, as the baby grows, eventually expand the tub.

Cloud Infrastructure Isn’t A Zero-Sum Game

Serverless isn’t here to destroy Kubernetes. The cloud infrastructure space race isn’t a zero-sum game. Kubernetes is an obvious evolution following OpenStack and can be run successfully inside of it. There will be OpenStack users for a long time to come, and there are also reasons many companies have moved on from there. Serverless is another tool in the belt of forward-thinking development teams. And increasingly, it can be run on top of Kubernetes (see Knative), enabling you to get the benefits of the simplicity of serverless and the complexity of Kubernetes where it makes sense for both in your stack. Eventually, we might even sort out distributed data, but that unicorn is a topic for another article.

Today, serverless-architecture can be done with Knative running inside of a cluster built and maintained by the company itself. But even if it’s a third-party service offered by a cloud provider, this isn’t really relevant, and the underlying technology is going to be Kubernetes in more and more places as the project continues to mature at breakneck speeds.

Serverless Is Just One More Tool

Serverless isn’t wiping away Kubernetes any more than containers got rid of operating systems. The two are going to be increasingly and intricately intertwined — iterative innovation made possible by standing on the shoulders of open-source giants. When teams sit down to architect what’s next, very few are going to go all-in on entirely serverless stacks, and very few should. Nonetheless, increasingly more companies are going to implement a part of their stack with serverless technology because there are specific problems for which it is unquestionably the best tooling around (occasional and fast-running workloads for example).

The trend of cloud-native open-source technology adoption will resemble a hockey stick, and we’re only just at the very beginning of the handle — or whatever metaphor means this market is exploding or, perhaps more accurately, just beginning to explode. Every year, there will be new technology that builds on what exists today. But you’ll keep hearing buzz about Kubernetes because it’s forming such an important foundation for everything coming above it.

Serverless isn’t the future. But serverless is definitely is a key part of the future.