The emerging practice of DevOps is upending longstanding infrastructure management models in the enterprise, and networking is at the center of this transition. In a world where everything is defined in software, provisioning networking resources becomes part of the app development and integration process. And aside from determining the initial requirements, this is largely left to automation.
This means enterprise networking specialists must not only acquaint themselves to new ways of doing things, but also to new tools with which to do them. And in large part, these tools will require broad reach into data center, cloud and edge topologies.
Cisco and Google have been working toward this goal for some time now, devising ways to formulate integrated application environments in the cloud using Google’s Kubernetes container management platform. Containers are considered a key asset in DevOps environments because they allow microservices to perform a wide range of functions for multiple apps, so coders don’t have to reinvent the wheel every time they need something to happen. The project is now mature enough, in fact, that the two companies are offering prizes totaling $160,000 to developers who create revolutionary apps on Cisco’s Container Platform with Google Cloud services.
Meanwhile, Red Hat has come out with the latest upgrade to its Ansible orchestration engine that allows organizations to support different vendors’ software-defined networking (SDN) operating systems, such as Cisco’s NX-OS, Arista’s EOS and the open-source VyOS. The idea is to provide a means to scale DevOps projects across multiple clouds and platforms, giving the enterprise broad flexibility to push applications over an increasingly interconnected digital ecosystem.
The symbiotic relationship between DevOps and SDN is a dream for developers and the enterprise in general, says Ramesh Ganapathy, assistant VP at software developer Mphasis. Not only does it provide an ideal environment for network operators to transition seamlessly from physical to abstract infrastructure, it enables uninterrupted digital experiences for users. By removing the pain points of configuring and upgrading network architectures, it paves the way for automated service delivery and, ultimately, the kind of elastic data environments that drive new services and new revenue streams.
Nevertheless, says F5’s Lori MacVittie, a number of myths have cropped up around DevOps and networking, none of which actually undermine their value to emerging business objectives but still should be acknowledged if the enterprise hopes to optimize its ROI. One is that automating tasks speeds up deployments. In reality, speed is achieved by eliminating wait times, which is best addressed by orchestration, not automation. As well, don’t expect automated processes to reduce human error to a significant degree; that can only come about by reducing manual inputs through strategic script analysis, code reviews and other means to cut “defect density.”
Undoubtedly, the enterprise will experience many fits and starts before it completes the transition to a DevOps-led digital business model, and much of this will center on getting the networking piece right. Since we are talking about software, expect the world of networking to become much more fluid and change-oriented, so that the process of perfecting it will go on forever.
Arthur Cole is a freelance journalist with more than 25 years’ experience covering enterprise IT, telecommunications and other high-tech industries.