Discover PerformanceHP Software's community for IT leaders // July 2014
DevOps for the real world
Author Gene Kim says regular companies, not just unique startups, can create high-trust organizations.
Gene Kim founded Tripwire, has written and contributed to numerous industry handbooks, authors a popular blog, and even co-wrote a novel that turned a DevOps transformation into material for compelling fiction. He has made it his mission to understand high-performing organizations, and to spread the word.
Lately, this front-line DevOps advocate has been countering the notion that DevOps and continuous delivery are only suited to superstar companies as rare and magical as unicorns. Not so, he says. As he gears up to host the DevOps Enterprise conference in San Francisco, Oct. 21–23, 2014, we asked him to explain why—and how—everybody else can turn their DevOps attempt into a success story.
Q: You’re well known for talking about "horses and unicorns." What are you referring to here?
Gene Kim: The unicorns are the classic DevOps success stories. But the problem is that if there’s anything horses hate, it’s hearing stories about unicorns. They hear about unicorns and think, "That’s not us—that won’t work for us. It’s easy for them because they were born that way. But we have decades of legacy systems, mainframes, COBOL…"
Q: They have a point, though, don’t they?
GK: What’s special about horses is that they often have decades of history of incredible economic success, so they’re not in a situation of existential risk. It’s not like, "If we don’t do this DevOps thing, the entire company fails." However, in many unicorn success stories, like at Amazon, Netflix, or Etsy, the future of the company was genuinely at risk and everybody knew that. They had to adopt DevOps work patterns. This is more difficult to do if you’re a large, successful bricks-and-mortar retailer, because "not doing DevOps" isn’t a clear and present danger. You can kick it down the road another year—until it’s too late.
It’s definitely harder to do these innovative, disruptive activities like DevOps in larger, more conservative organizations. They’re more likely to have a low-trust, command-and-control bureaucracy. However, despite all these cultural impediments, there are horses who have overcome them and achieved unicorn-like levels of performance.
Q: Why is a high-trust environment so important in DevOps?
GK: Jez Humble and I worked with Puppet Labs over the last two years to understand what makes great IT organizations great. This year, among 9,600 organizations, we found that one of the highest predictors of IT performance and organizational performance—which is a proxy for competitiveness—was whether it was a high-trust or low-trust environment. In low-trust organizations, messengers of bad news are shot. New ideas are crushed. In a high-trust environment, bridging between functions is encouraged. New ideas are encouraged.
We know one of the classic hallmarks of high-performing DevOps teams tends to be small teams with high trust. There’s no doubt in my mind that one of the biggest management challenges of this decade is how to turn low trust into a high-trust model.
Q: What does that look like? Can you give us an example in which a high-trust organization will react differently than a low-trust organization?
GK: I’m just now rereading a book by Steven Spear called Chasing the Rabbit. He’s famous for writing Decoding the DNA of the Toyota Production System. He says we used to think that management was all about making the correct decisions. He asserts that, instead, management’s most important job is to foster a culture that can turn local discoveries into global changes. In successful organizations, when something goes wrong, those two questions are: "What got in your way?" and "What can we do so it doesn’t get in your way tomorrow?" That’s high trust. In a low-trust organization, you just fire somebody.
In the IT world, here’s how it works: If people are continually asking questions about what caused problems, and management is supporting fixing those problems by allowing technical debt to get paid down, we expect reliability to go up. Furthermore, we expect flow to increase, too. We’ll see faster time-to-deploys, measured from when code is declared dev-complete, checked into version control, tested successfully, and deployed to production. In a high-trust organization, these things will be measured in minutes or hours—instead of days, weeks, or months.
Q: You encourage the "horses" to believe that they can be successful in implementing DevOps. So what’s the dose of reality that needs to go with that encouragement?
GK: When people ask how much time is required for these transformations, the answer seems to be about three years, whether horse or unicorn. Because these transformations are an innovative act—in other words, they’re something the organization hasn’t done before—it’s a foregone conclusion that the first attempt isn’t going to work. So you need the permission to fail and the relentlessness to keep pursuing it.
Q: Executives outside of IT might not like to hear that. How do CTOs find that wedge issue in their organization that will help them convince decisions makers to pursue DevOps?
GK: There are patterns that are emerging. One is that in almost any organization there’s a constant pressure to reduce time to market. So you need to be someone who can connect the dots and say, "We need to reduce time to market, and DevOps is a way to do that." That’s the biggest bottleneck in most value streams. Another way is to say, "We need to develop this ability to outperform the competition in the marketplace." And show how DevOps can do that. Then find that project that will have a large enough impact that it can create a material contribution to the business, but is small enough that you don’t jeopardize the organization if you screw up.
Gene Kim (@realgenekim), founder and former CTO of Tripwire, has worked with leading Internet companies on improving deployment flow and production rigor around their IT operational processes. He is the co-writer of The Phoenix Project: A Novel About IT, DevOps, and Helping Your Business Win.
Register for the DevOps Enterprise conference in San Francisco, Oct. 21–23, 2014.
HP Software’s Paul Muller hosts a weekly video digging into the hottest IT issues. Check out the latest episodes.
Introduction to Enterprise 20/20
What will a successful enterprise look like in the future?
Challenges and opportunities for the CIO of the future.
What the workforce of 2020 can expect from IT, and what IT can expect from the workforce.
Data Center 20/20
The innovation and revenue engine of the enterprise.
Dev Center 20/20
How will we organize development centers for the apps that will power our enterprises?
Welcome to a new reality of split-second decisions and marketing by the numbers.
IT Operations 20/20
How can you achieve the data center of the future?
Preparing today for tomorrow’s threats.
Looking toward the era when everyone — and everything — is connected.