Mirantis announced it has joined Airship, an open source project for automating the provisioning of clouds, as part of an effort to accelerate deployment of Kubernetes on bare-metal servers that AT&T will employ to drive 5G networking services.
Airship was founded by AT&T, Intel and SKT to marry a cloud management framework based on OpenStack to Kubernetes.
Mirantis CEO Adrian Ionel says Mirantis has been working with AT&T for more than 18 months on deploying instances of Kubernetes based on a distribution curated by Mirantis, which are being deployed on bare-metal servers to drive the delivery of next-generation 5G services. Most of those 5G services are too latency-sensitive to tolerate the deployment of a virtual machine underneath those instances of Kubernetes.
Mirantis, as a member of AirShip, is being tasked with streamlining the configuration process for deploying Kubernetes in an on-premises environment, providing support for multiple operating systems on which Kubernetes can deployed and integrating Ironic, a tool for provisioning OpenStack on bare-metal servers with Drydock, a declarative tool for provisioning hosts that was developed under the auspices of the Airship initiative.
Containers and Kubernetes make it possible for carriers such as AT&T to employ container network functions (CNF) as an alternative to virtual network functions (VNFs) that are tightly coupled to a specific instance of a virtual machine. In contrast, CNFs are more portable across different distributions of Kubernetes, assuming that no proprietary extensions have been made to the version of Kubernetes made available by the Cloud Native Computing Foundation (CNCF).
In general, tier one carriers such as AT&T are much further along in terms of deploying Kubernetes on bare-metal servers than the average enterprise IT organization. Most enterprise IT organizations today prefer to deploy Kubernetes on top of a virtual machine out of concerns of being able to isolate instances of Kubernetes on a physical server and the fact that most of them lack the tools and expertise required to deploy Kubernetes on a bare-metal server.
But applications deployed at the network edge generally need to support applications that can’t tolerate any more than 20 milliseconds of latency. To meet that goal, providers of 5G services need to streamline the stack of software deployed at the network edge as much as possible. Carriers also need to work through all the issues associated with deploying Kubernetes at scale across what will eventually become hundred of thousands of instances of Kubernetes deployed at the network edge. Carriers are already starting to deliver 5G services in a limited number of markets. But rolling out those 5G services at scale is in many cases contingent on how quickly they can master all the intricacies of Kubernetes.
In the meantime, it’s clear everyone from enterprise IT organizations to IT services providers will be watching the progress AT&T makes deploying Kubernetes on bare metal servers. After all, it’s only a matter of time before most of them follow suit.