The world’s biggest tech companies use their software development talent to automate data centers to the extreme, taking a holistic platform view of their infrastructure rather than focusing on one stack or one location at a time. They are what Marshall Van Alstyne and his co-authors refer to as “platform companies” in Platform Revolution, a book about the way Facebook, PayPal, Alibaba, Uber, and others have disrupted entire industries by harnessing the “platform business model.”
A platform marketplace, which connects sellers and buyers at massive scale, is underpinned by an infrastructure platform, and that platform is impossible to achieve without extreme automation, enabled by software. That means successful infrastructure and operations teams of the future will have software developers not only work side-by-side with operations staff but become fully integrated members of enterprise infrastructure and ops teams.
That’s according to Gartner’s Milind Govekar and Dave Russell, who in the opening keynote of the research firm’s annual IT Infrastructure, Operations Management, and Data Center Conference in Las Vegas Monday shared Gartner’s ideas about the way forward for enterprise IT teams. One key idea: hire programmers.
“We will have to start recruiting programmers in I&O (Infrastructure and Operations),” Govenkar, a research VP, said. “We will no longer have administrators only in I&O.” This needs to be done in order to build agility and innovation into the company infrastructure and “clearly will have a big impact on skills.”
Using as an example the technology team structure of one of Gartner’s clients, whom they did not name, the analysts said a good place for the modern IT and ops team would be to act as an operations and service management layer of a “business embedded operating model,” sitting between a central infrastructure team on the bottom – which oversees the nuts and bolts of infrastructure as well as infrastructure automation and, importantly, spends a lot of its time writing APIs for using that infrastructure – and a platform engineering team on top, which delivers the platform to various business units:
It’s no secret that the old approach to staffing and structuring teams that build and manage enterprise technology infrastructure is on its way out. Companies cannot take full advantage of modern software capabilities if they have a dedicated infrastructure stack for every app, managed by system administrators conditioned to resist change by decades of putting risk reduction above all else.
Companies like Facebook or Uber didn’t have to deal with legacy. That’s one advantage of building an application that’s not like anything that ever existed before. Automation enables a single Facebook data center tech to manage 25,000 servers, and in some cases up to 40,000 servers. The company built this kind of automation out of necessity: it’s necessary for the kind of scale and agility its business requires.
Unlike traditional enterprises, platform companies never stop changing their applications, rolling out updates and new features often on a daily basis. Their infrastructure needs to be flexible enough to accommodate that while being able to reach users globally.
But traditional enterprises are increasingly starting to behave like platform companies. Business leaders want to roll out new software capabilities continuously to enable delivery of new products, new employee services, or even new business models. Enabling that agility on the IT and ops side means applying software development tools and methods – such as automation, versioning, APIs, immutability, and agile techniques – to IT infrastructure management.
In Russell’s words, “Stop building systems to last and start building systems to facilitate change.”