Serverless on Kubernetes: A Real-World IoT Production Scale Implementation Without Vendor Lock-in

Serverless computing provides an abstraction layer on top of existing computing environments where the unit of deployment consists of a single function. Abstracting away infrastructure offers important advantages, specifically for improving deployment.

In this talk we introduce serverless architectures built on top of Kubernetes in order to prevent lock-in from cloud providers’ proprietary offerings.

Kubeless is one of the most promising framework for fulfilling these requirements. We describe how we used it in an architecture for a large Internet of Things (IoT) digital farming use-case to collect and analyse data. A deep dive into development, deployment and runtime will be presented.

Objective of the talk

To understand how a FaaS (Function as a Service) model can be built on top of Kubernetes and provide an overview of how this was used in a practical IoT scenario. Advantages and disadvantages of an open source in compare to a vendor proprietary FaaS-system derived from a real world use-case.

Required audience experience

A general understanding of containers, Kubernetes and Serverless will help to follow the talk.

You can view the slides below:

Raphael Herding and Grischa Ekart – Serverless on Kubernetes

Track 3
Location: Burton & Redgrave Date: November 7, 2019 Time: 4:40 pm - 5:25 pm Grischa Ekart Grischa Ekart Raphael Herding, Bayer AG Raphael Herding, Bayer