A couple of years ago, container networking typically consisted of doing basic port mapping on a single docker host, putting in place a bit of DNS config and then hoping for the best. However, as use of containers has moved from developer experiments through to real production deployments, there have been pressing requirements on the container ecosystem to develop “real” multi-host networking technologies.
In this 40 minute session, we will look back at the evolution of container networking from those initial port mapping solutions, through initiatives such as Docker’s libnetwork and the Mesos net-modules framework, through to the standardization of container networks with the CNI (Container Networking Interface). We’ll explain how these fit into various orchestration frameworks and cluster managers like Kubernetes and DC/OS. We’ll also explain why networking is not just about getting packets from A to B - it’s also, critically, about ensuring that only the right (!) packets get from A to B based on the security policies put in place for your containerized services and applications.
With this in mind we’ll explain how projects such as Canal are working alongside various communities to provide fine grain policy support in Kubernetes and Mesos.
No specific pre-requisites, although a basic knowledge of container technologies and networking might be helpful.
Understand the current state of container networking, how we got there and what the future holds.
// Karthik Prabhakar
is Director of Solution Architecture at Tigera. He is focused on growing the adoption of new networking and policy capabilities available with container frameworks and cloud infrastructure. In the design of cloud-native networks, he leverages his years of experience at Red Hat architecting provider-scale and enterprise cloud solutions along with best practices and lessons learned from other Chief Architect and Principal Network Engineer roles.