Function-as-a-Service on Kubernetes using Knative

The CERN Cloud Infrastructure team provides compute resources as a service to teams 
across CERN. Users can provision resources to process experiment data, host web 
applications, and accomplish other computing tasks. 

The CERN cloud is built using OpenStack, an open-source suite of technologies for 
provisioning and managing a variety of computing infrastructure services, including: 
virtual machines, “bare metal” machines, container orchestration, storage, 
authentication, and networking. For container orchestration, we use OpenStack 
Magnum, which supports multiple container orchestration engines, such as Kubernetes. 
Kubernetes is a widely-used platform for deploying and managing applications in 
containers. Our team uses these technologies in part because they are open-source and 
developed by active communities with strong ties to information technology industries. 

Our team is interested in extending our infrastructure offerings to include serverless 
application deployment, which abstracts away even more infrastructure management 
from the user than traditional services like virtual machines. Amazon Web Services 
(AWS) popularized the notion of serverless computing with their introduction of AWS 
Lambda, which provides a Function-as-a-Service (FaaS) platform. Our team strives to 
offer a similar serverless computing service. We are particularly interested in creating a 
FaaS platform on top of Kubernetes, so that we can simultaneously reap the benefits of 
using a container orchestration engine. 

Knative is a relatively new technology that extends Kubernetes to support serverless 
computing patterns. It is also an open-source project backed by major industry players 
like Google, IBM, and Red Hat. For these reasons, Knative is a natural candidate for our 
purposes. My project was to investigate Knative’s functionality and how our team could 
use it to add a FaaS platform to our suite of existing cloud services.