Rather than rely on virtual switches and network overlays to connect containers across networks, Metaswitch, a provider of networking software, has been making the case for a Project Calico effort that relies on the Border Gateway Protocol (BGP) to network containers such as Docker.
Today the team that leads the development of Project Calico announced that it is now adding support for the Container Network Interface (CNI), which is an open interface for configuring container networking supported by the open source Kubernetes framework for orchestrating containers, which is spearheaded by Google.
Christopher Liljenstolpe, director of solutions architecture at Metaswitch and evangelist for Project Calico, says the trouble with virtual switches and network overlays is that they take too long to configure. Developers often only need to spin up a container for a few hours. They don’t want to have to wait days, sometimes even weeks, for an IT staff to configure a virtual network using technology that was originally developed to support virtual machines.
Integration with Kubernetes, says Liljenstolpe, will make it possible for developers to network containers on their own using BGP gateways that are familiar to most IT networking staffs. The end result is that developers become more efficient using a model where every container winds up with its own IP address, says Liljenstolpe.
That’s significant, adds Liljenstolpe, because going forward it means that IT organizations will be able to apply policies to each individual container each time one is created by a developer.
While it’s clearly still early days in terms of adoption of containers in the enterprise, it’s already clear that networking is emerging as a major bottleneck when it comes to fostering IT agility. As increased developer productivity and IT agility are the two primary reasons for adopting containers in the first place, legacy approaches to networking in the enterprise are not conducive to achieving either one of those goals.