This year’s State of DevOps Report, which has become a standard for defining progress and trends in the software development and delivery community, is being published today. Presented by Puppet and DevOps Research & Assessment (DORA), this year’s report takes on some new areas of research, including the role that transformational leadership plays in what traditionally has been a grass-roots movement, and the impact DevOps has on multiple organizational objectives—not just the bottom line.
Now in its sixth year, the 2017 report features the responses of more than 3,000 participants around the world, 27% of whom work on DevOps teams. That’s up 11 percentage points from 2014, when just 16% of respondents were part of those teams. That increase is an acknowledgment that DevOps works and that strategies are afoot for shifting entire organizations from older ways of working to newer DevOps processes.
As that transition takes place, DevOps organizations may realign in some key respects to drive high IT performance, even as the measurements for organizational performance success extend beyond profitability-related goals.
These five key highlights from the 2017 State of DevOps Report speak to these issues.
1. Good management has an amplification effect
The DevOps community has sometimes been guilty of maligning leadership for restraining efforts to make changes to improve IT and organizational performance. “You rarely hear anyone talking of the influence strong leaders have over DevOps transformation and IT performance,” says Alanna Brown, director of product marketing at Puppet. She is a co-author of the report, along with Nicole Forsgren, CEO and chief scientist at DORA; DevOps researcher and author Gene Kim; Jez Humble, co-founder and CTO at DORA; and Nigel Kersten, chief technical strategist at Puppet.
It’s there, though, with high-performing IT teams reporting that they have leaders with the strongest behaviors across the five dimensions of vision, inspirational communication, intellectual stimulation, supportive leadership, and personal recognition. Teams that report having the least transformative leaders in these respects were statistically less likely to be high IT performers.
Transformational middle management and C-level leadership can take DevOps from pockets of success to organizational success by driving the widespread adoption of key technical practices and new cultural norms for continuous delivery. “We can see grass-roots efforts succeed and DevOps practitioners achieve transformation on their own, but if you have good management in place, that bolsters and amplifies these effects,” Forsgren says.
But the survey findings also show that transformational leadership in the absence of suitable architecture, good technical practices, use of lean management principles, and other factors is not enough to achieve high DevOps outcomes, says Kim.
2. Higher IT performance depends on strong continuous delivery practices
The research found that high performers deploy code 46 times more frequently than their low-performing peers, and their lead time required to deploy changes into production was less than one hour. The survey results confirm the importance of architecture in achieving high IT performance in a continuous delivery setting: Loosely coupled services, along with loosely coupled teams, are the biggest contributors to successful continuous delivery.
“Working in small batches and structuring teams so that they can independently deploy the services they’re responsible for without a lot of coordination and hand-offs enables higher throughput, along with higher quality and stability,” Brown says. Teams that can decide which tools they use do better at continuous delivery because they can make these choices based on how they work and the tasks they need to do. Transformational leaders tie into this, Brown adds, because they “can empower their teams to choose the tools that they want to use, as well as to figure out how best to structure the team to deliver smaller units of work.”
Following up on last year’s findings that trunk-based development is an important factor for better continuous delivery than is leveraging long-lived feature branches, the report this year turned its attention to just how long branches and forks can live before being integrated into trunks to contribute to better software delivery performance. “We found that high performers have the shortest integration and branch lifetimes, typically lasting hours,” Forsgren says, while low performers take days. “Teams should ask themselves where they stand and start to work on this,” she adds.
3. Higher organizational performance depends on strong lean product-management practices
Having stronger capabilities in the IT practice of lean performance management—which enables a feedback loop with customers to quickly develop and deploy the features they want—drove overall organizational performance and contributed directly to the bottom line for survey respondents, according to the report.
“When you can gather and quickly test feedback from customers in things like A/B testing, you can make small changes quickly and take an experimental approach to product development,” says Forsgren. “We see this happening with the strongest and most innovative companies we have out there today.”
4. Automation is critical, even when the going seems slow
Despite lingering distrust of automation and a persistent belief that it is a job-killer, high performers automate significantly more of their configuration management (33% more), testing (27% more), deployments (30% more), and change approval processes (27% more) than do low-performing teams, the research finds. The reward is that DevOps teams actually get to do more interesting, strategic, and innovative work, Brown says.
But while medium performers are deploying software faster and more frequently than are the low performers, they are actually less automated in terms of deployment and change-approval processes compared to the low-performing group. Medium performers handle 47% of deployment work manually, compared to 43% of low performers, and handle 67% of change approvals manually versus 59%. Although medium performers have automated enough of their work to see IT results, they’re uncovering mountains of technical debt and grappling with how to address that.
“The general rule is that things get worse before they get better,” Kim says. “They’re finally uncovering daily workarounds and problems hidden in the status quo. They couldn’t solve what they couldn’t see before, and as they see more, there’s more to solve, and that’s the start of the journey.”
The authors’ advice to those organizations is to recognize that it’s going to be hard for a while, as they burn down the technical debt. But once that’s finished, further automation is achievable. “So long as you continue the transformation journey, you will get through it,” Forsgren says.
5. DevOps is about more than meeting financial targets
This year, high performers were twice as likely to achieve their own reported goals across both financial and nonfinancial measures. High performers in both for-profits and not-for-profits were twice as likely to achieve or exceed objectives in areas ranging from operating efficiency to customer satisfaction to the quality of products or services delivered.
“An important new finding,” says Kim, “is that DevOps helps with the achievement of all organizational objectives, whatever type of organization it is”—government or charity, for example. Similarly, it helps those working in for-profit organizations to see the relevancy of DevOps throughout the organization, to efforts in corporate social responsibility, for example.
“For so long, we’ve been hearing people in other organizations say, ‘What about us, if we work in government or non-profit or in a corporate social responsibility area?’” she says. “So this year we added measures to investigate these other types of organizational performance for all types of organizations.”
Whether an organization is trying to generate profits or not, today it depends on technology to achieve its mission and provide value quickly, reliably, and securely to customers or stakeholders. And how well your technology organization performs will predict overall organizational performance.