Digital transformation lessons from 3 DevOps leaders

In just a few years, DevOps has evolved from a grassroots effort aimed at transforming the way systems are developed and deployed in early-adopter organizations into a growing mainstream practice. The realization that accelerated demand for technology services can no longer be met with traditional methods has led to increased adoption of DevOps methods across many companies—part of a broader digital transformation roadmap.

“In order to survive in the marketplace today, you need a world-class technology organization,” says Gene Kim, founder of IT Revolution, which co-produces the DevOps Enterprise Summit events. “And world-class technology operations embrace modern technical practices.”

With its promise of fast, frequent, high-quality releases, DevOps has become the primary vehicle for driving digital transformation, says Robert Stroud, principal analyst at Forrester Research. Companies across most industries are now using it to deliver enterprise systems, Internet of Things technologies, mobile apps, and e-commerce functionality, according to Forrester’s 2017 DevOps heat map.

But pioneering IT leaders in a few industries are out in front of the competition, thanks to their early embrace of DevOps approaches. And the benefits are clear.

DevOps leaders are two to three orders of magnitude more productive than mainstream IT organizations, according to IT Revolution’s benchmarking data from 27,000 respondents over the last four years.

“In an age where how you acquire customers and how you deliver value to them are the key competitive differentiators, those companies that are the most productive win,” Kim says. What’s more, IT functions that exist further along in those companies' DevOps programs are hiring more than their competitors. “Increasing head counts and budgets are indications that DevOps creates real organizational value,” he says.

The early adopters

The first DevOps poster children were digital powerhouses and startup unicorns such as Amazon, Etsy, Facebook, and Netflix. Retailers such as Nordstrom and Target, facing a marketplace turned inside out by digital technology, were not far behind. Other early adopters were financial services firms fending off FinTech upstarts, utilities and telecom providers rethinking their business models in a commoditizing market, and software makers transforming themselves for the as-a-service world.

DevOps front-runners have certainly sped up production. Today, Amazon, for example, deploys code an average of every 11.7 seconds, and Netflix some thousands of times per day. But more importantly, they’ve developed platforms that can deliver the kind of rapid innovation, quality development, and customer value that’s required in a business environment marked by more frequent shifts in technology and demands.

From movement to mandate

Some 90% of organizations are implementing, expanding, or planning to implement DevOps, according to Forrester’s Stroud. It’s not a fringe activity—it’s a competitive mandate. But just 13% of companies have fully implemented DevOps. Those select few are the DevOps first movers.

Capital One

Capital One’s IT transformation began six years ago, driven by a “recognition that the winners in banking will have the capabilities of a world-class software company,” says Tapabrata Pal, director and engineering fellow at the financial services firm. The need to rethink systems development was clear. “In the banking sector, speed is the new currency,” says Pal. “To remain competitive, you need to react fast to customer demands and a fast-changing technology space. DevOps practices help you get there.”

The engineering team pursued increased automation with a simple goal: continuous integration for new projects. Prior to that, everything, from the code-merging process to library management to basic testing, was manual. DevOps practices were just emerging, but they aligned with the bank’s goals. Now, says Pal, “the early adopters like us are ahead in the game.”

KeyBank

The introduction and expansion of DevOps philosophies and practices have dramatically transformed the IT function in several large financial services firms. In the past, attempting to introduce new systems was an exercise in frustration, says Christopher McFee, manager of DevOps engineering at KeyBank. “Buildup of manual workflow, fear of change to [proven] processes, and the confusion with the number of communication points create a lengthy, error-prone delivery process,” he says. Legacy systems, acquired complexity, and a myopic focus on cost-cutting also held financial services back.

Around 2015, KeyBank leaders recognized the competitive advantage that streamlined delivery of new systems would provide. They began using the practices to deliver new online banking systems, speeding up infrastructure delivery through containers, increased automated testing, and continuous deployment pipelines. “We had a lot of great results with our initial delivery, and now we are trying to make sure we are meeting the demand for many of our other delivery efforts,” McFee says. “In order to overcome some of those challenges, we are focusing on collaboration across the entire organization and making the work visible.”

Built for innovation

Bureaucracies, says Kim, are designed to preserve the status quo. In times of extreme disruption, companies need to enable risk-taking and innovation, “which is absolutely what DevOps is,” he says. “DevOps can enable people to do things in different ways and avoid complacency.”

DevOps leaders have found that the approaches enable the organization not only to go fast, but also to experiment with greater ease, says Forrester’s Stroud.

Disruption at Dynatrace

That's been the case for Dynatrace, a software manufacturer that decided to disrupt its own business model in 2011. The company saw the shift in the enterprise software market and wanted to be at the forefront of it, Dynatrace CTO Bernd Greifeneder told Stroud. To speed up innovation as it moved its digital performance management product to an as-a-service model, the company accelerated its transition from DevOps newbie to leader, with a focus on continuous experimentation and improvement.

It began handling infrastructure as code, enabling it to focus on delivering increasing application value to customers. In what Stroud calls a key signal of DevOps' maturity, the company now monitors product performance and usage in real time and adapts its products to customer needs, addressing problems before they happen.

At Capital One, soup-to-nuts product development ownership has yielded significant value for customers and the company. Previously, Capital One developers built software and turned it over to quality teams to iron out the bugs, and then the quality teams turned it over to operations, which sent it out to production and would support it going forward.

Now a product team continuously builds, tests, and deploys its software, using containers and microservices. They push the apps in the public cloud and support production, too. “[We] can quickly resolve the issues that come up,” says Pal. “It also changes the mindset of developers about the importance of quality, of code delivery, and what it means to develop code for running at scale with high reliability in an environment geared toward resiliency.”

The approach has helped the company innovate apace. Last year, Capital One became the first bank to enable customers to access their financial accounts through Amazon Alexa-enabled devices. This year, it was the first US bank to introduce a natural-language SMS chatbot. The DevOps move has also enabled the company to harness “fast data” for fraud detection. “DevOps helped us claim victories in many instances,” Pal says. “It’s a powerful way of working.”

Insider tips: How to follow the leaders

It’s no longer a question of whether DevOps can create business value, but rather one of how to deploy it more broadly in the enterprise. “We are seeing it take hold in large enterprises in retail, financial, insurance, airline, and media companies, just to name a few,” says KeyBank’s McFee. “Taking a DevOps approach enables all industries to deliver value to their clients quickly—whether it’s new features, increased stability, or quality. It’s becoming a requirement for all industries.” And while DevOps pioneers may be further along in their transformations, their early efforts reveal insider tips for others:

• Socialize DevOps. Capital One pursued several avenues to spread the gospel of DevOps—building communities of practices, setting up shared services, hosting hackathons, meetups, and office hours. It even put on an internal software engineering conference for thousands of employees, with presentations and booths that highlighted DevOps work.
• Reconsider sourcing strategies. Partnering with experienced third-party providers can play a critical role in delivering technology-enabled business value, but full-scale outsourcing can hold companies back. Organizations that are highly outsourced contract “for stasis in multi-year chunks,” says Kim. “They tend to fare more poorly in times of incredible change.”
  
  Capital One evolved from a mostly outsourced to a mostly insourced organization during its DevOps transformation, co-locating engineers with the business based on the products they build. “We are also moving these operations upstream and integrating them into development teams,” says Pal.
• Rework processes and responsibilities. Process automation is the DevOps foundation. But DevOps pros first remove redundant processes. “Every traditional process has significant delays,” says Stroud, who suggests applying value steam mapping to detect superfluous practices.
  
  DevOps masters have also freed fellow team members from legacy roles and fully aligned them around services or applications. They have also given them space to grow into their new roles. “Management needs to allow time for the team to learn,” Stroud says. “There will be some initial challenges. Feedback should be used to refine and enhance processes, not add additional checkpoints.”
• Build your own. There is no one-size-fits-all approach to introducing and expanding DevOps practices. Pal strongly believes in BYOD: building your own DevOps. “Learning from others and listening to experts are helpful,” he says, “but every DevOps implementation is different. When you start building your own, you need to start small—measure, learn, correct, and expand the scope.”
• Don’t wait. There is enormous value in simply getting started. “Just try something and gather as much feedback as you can,” says McFee. “If something isn’t working, experiment and iterate on the process.”
  
  The sooner you begin, the better. Being a follower may once have been a safe bet, but today the real risk is not taking the leap. “At a certain point, you’re no longer doing it to win,” says Kim. “You’re doing it just to not lose."