Author: William Chia

  • Predictions 2020: How to Make the Case for Adopting CI/CD in Your Organization

    Predictions 2020: How to Make the Case for Adopting CI/CD in Your Organization

    The practice of continuous integration and continuous delivery (CI/CD) is fundamental to modern software development. And yet, many organizations are still using slow, error-prone, manual processes to build, test and deploy their software. Sometimes, the blocker can be engineers who don’t want to change how they’ve always done things, but most often folks in the trenches are chomping at the bit to automate. All too frequently, it is IT leadership and executives who need to be convinced that making a large investment in updating tools, training and culture will net returns greater than the cost.

    If you want to change management’s mind about an investment in CI/CD, focus on making a strong business case. This may sound daunting, but it’s easier than you think. Learning a bit about how to advocate for budget and organizational change can go a long way. Just think about what life would be like if, instead of sitting in endless change management meetings, you could get your code to production simply by merging to master while automation takes care of the rest.

    This article will get you started with some concrete tips you can use to get CI/CD green-lit by your leadership.

    Start Small

    Agile and DevOps are all about breaking things up into smaller chunks. It may seem counterintuitive, but deploying more frequently equates to less risk because small changes are easier to orchestrate, troubleshoot and roll back than large complicated deployments. In that same way, you can reduce the risk to the organization by taking a DevOps approach to transformation. Rather than trying to upend your entire process, see how you can break the change up into small, iterative chunks.

    Almost every large enterprise that has successfully adopted wide-spread CI/CD started with one lighthouse team. Changing everything is a tough sell. Instead, approach management with a low-risk proposal: “We’d like to run an experiment to see if we can deliver more quickly and reduce costs. We have one team that will take one small project and test CI/CD automation to see what the results will be. We can leverage open source tooling to get started at a low cost.” Even the most stingy budget holder should find the proposal intriguing.

    Measure the Results

    Once you’ve started to conduct your experiment, it’s important to measure what you observe. What are the differences between the old way and the new way? The more concretely you can quantify it, the better. Saying, “We went faster,” doesn’t carry a lot of weight. But if you can quantify it, your message becomes more potent. For example, “Our previous cycle time from the moment a developer begins work on a ticket to the time that code is running in production and useable was 30 days. By automating some of our manual efforts using CI/CD we have cut that to five days.”

    For this part of the process there’s no need for exact numbers. Most likely if you aren’t doing CI/CD today, you likely don’t have the instrumentation in place to accurately measure cycle time. But don’t let that be a blocker. You can give estimates. “From about one month to one week,” is still a powerful story.

    Some other metrics to capture that can help make your case are:

    • Number of tests run.
    • Code coverage in terms of percentage of code tested.
    • Number of people-hours required.
    • Number of defects in production or number of regressions/roll backs needed.
    • Mean time to repair (MTTR): The average time from when a bug/failure is discovered in production to when the fix is running.
    • Frequency of releases—how often new capabilities are available for users.
    • Cost to deploy: This is a very powerful metric if you can estimate it. A good rule of thumb is: (number of people minus hours, times average hourly rate) plus (cost of assets and services required).

    Aim to use three to five metrics to prove your case. You don’t want to overwhelm with data. The point is to show that things are better with CI/CD, not that you’ve gathered a lot of numbers. Use a simple before-and-after format to show the change. State the changes in terms of increased value, reduced cost and/or reduced risk. For example, “We increased value for our users by shipping 10 times more features in the same time frame, reduced deployment costs by 80% and reduced risk by increasing code coverage by 30 points from 40% to 70%.” 

    Couple Your Efforts to Business Outcomes

    A critical element of IT success is how closely you can partner with the business. It can be next to impossible to get initiatives green-lit if the only benefit is to make things better for IT. But if you can show how some extra effort spent on IT modernization can drive results for the business, then you’ll find you start getting enthusiastic approval from leadership.

    Collaborating with people outside of IT or software R&D is a career-enhancing skill every technical practitioner should strive to improve. 

    Get to know what efforts in sales, marketing and product are dependent on IT. For example, the sales department may desperately need additional capabilities in their forecasting tool if they are going to execute strategic deal acceleration, ensure quarterly numbers are consistent and avoid unpredictability with shareholders. In other words, the fact that your slow delivery cycle can’t get those new features on the release calendar for another three months is putting the stock price in jeopardy.

    To avoid this tragedy, you can tie your CI/CD efforts to the sales outcomes. By getting to know the pain points of your sales colleagues, you can craft a powerhouse “ask” to leadership. “We’d like implement CI/CD to automate deployments for our sales data management system. This will allow us to implement new filters in a matter of days rather than months which sales needs to hit their numbers.” 

    Pulling It All Together 

    It can be demoralizing to feel like you are behind the curve. Looking at other businesses who have already adopted CI/CD it can feel like you might never catch up. If this is the case, some axioms are appropriate, such as the famous Chinese proverb from the Tao Te Ching “A journey of a thousand miles begins with a single step.” And as Theodore Roosevelt said, “Comparison is the thief of joy.”

    The reality is that you are not alone. Many organizations, especially large enterprises, are relying on outdated software practices. Convincing leadership to green-light IT modernization efforts can be a challenge. Adopting CI/CD in a meaningful way does require significant investment, but there is a lot you can do to smooth the path for them and earn their support.

    Begin by identifying a small adventurous team who can serve as an experimental testing grounds. Not only will their success help to convince leadership, but their example serves to light the way for other teams to adopt DevOps. Identify metrics you want to capture and be sure to instrument so you can measure the improvement. Pick a few metrics that show increase in value, decrease in cost and decrease in risk. Finally, reach across the org to collaborate with your non-engineering peers. By showing how your modernization effort drives outcomes for the business you’ll be sure to earn respect as well as additional budget to expand your CI/CD efforts.

    Want to learn more about what to expect in 2020? Join us Jan. 23 for our Predict 2020 Virtual Summit  featuring discussions from some of the industry’s best and brightest offering up their visions for the future. Sign up today for this free daylong virtual event.

    — William Chia

  • Digital Transformation Technology and Process: More Similar Than You Think

    Digital Transformation Technology and Process: More Similar Than You Think

    Right around this time of the year, the gym starts to get a little less crowded. Throughout January, it’s packed to the gills with folks freshly motivated by their New Year’s resolutions to get in shape. But by mid-February, the commitment seems to have all but dissipated.

    Why is it so hard to just commit?

    During my last workout, it occurred to me that the set of vacant treadmills next to mine was eerily reminiscent of the state of enterprise DevOps. According to Forrester, 50 percent of digital transformation efforts stalled in 2018. It seems commitment and follow-through are difficult in both personal and professional arenas. Admittedly, adopting and scaling DevOps is no easy feat. While reflecting on what does work, I found some parallels between the transformation process and DevOps itself.

    What Success Looks Like

    According to DORA’s “2018 State of DevOps report,” organizations with the highest level of DevOps execution, as measured by throughput and stability, also achieved the highest level of business performance in terms of profitability, market share and customer satisfaction. In a true case of going fast without breaking things, elite performers realized impressive performance gains:

    • 46 times more frequent deployments (throughput)
    • 2,555 times shorter lead times (throughput)
    • 7 times lower change failure rate (stability)
    • 2,604 times shorter time to recovery (stability)

    These metrics underlie a key tenet of DevOps transformation, which is breaking up large chunks of work into many small pieces.

    Shrinking the Change

    Smaller changes mean getting work out faster, with less risk. They take less time to plan, less effort to execute and are easier to troubleshoot. But, smaller code changes alone aren’t meaningful unless they are coupled with shorter release cycles. In essence, agile without CD doesn’t work.

    Stockpiling a few months worth of agile code iterations into a single monolithic release reintroduces an enormous amount of risk and negates much of the work done to adopt agile software development in the first place. Big queued-up releases often mean someone is stuck working the weekend to iron out all of the things that inevitably go wrong and no one should be stuck with weekend release duty. As Jez Humble put it, “The most important metric for continuous delivery is the number of person hours spent outside of business hours doing releases… and the goal for that is zero.”

    In practice, CD is what brings agile development to fruition; it’s why Gartner finds that 40 percent of IT departments have already invested in CI/CD with a full third planning to invest in 2019. CD is all about small pieces. In the same way agile enables smaller code changes, CD enables smaller releases.

    Not Just Smaller, but Iterative

    At GitLab we use the phrase, “just commit” as a shorthand for continuous delivery and deployment. The idea is that when you have a completely automated CI/CD pipeline, all a developer needs to do is commit code to their branch, and everything else needed to get that code running live in a production-like environment—from containerization to security testing—is automated. Similarly, getting code to production is as easy as merging the feature branch back into trunk with the entire production deployment automated. The message to the developer is, “Just commit your code and let your DevOps tooling do the rest.” This is CD in a nutshell.

    With this type of automation in place, the cost of making a change significantly decreases. This means the cost of getting started is very low, but so is the cost to keep going. Instead of spending a massive effort to map an entire path to your solution, you simply need to figure out your next step and keep iterating. According to GitLab’s “2018 Global Developer Report,” 67 percent of developers report the top challenge to getting work done is unclear direction. By shrinking the change, business leaders will find it easier to set direction for the next step without having to map the entire path. This brings us back to digital transformation.

    Commit to Digital Transformation

    Just like a monolith being decomposed into microservices, the practice of breaking large chunks into small pieces applies to driving digital transformation as well. Like Creighton Abrams said, “When eating an elephant, take one bite at a time.”

    How can enterprises shrink the change?

    For most organizations that achieve successful digital transformation, the initiative usually starts with a single team or innovation unit. This group proves the success for the rest of the organization. It develops that culture, tools and processes that other parts of the organization can then adopt. One example of this principle in practice is the transformation that Chris Hill, Head of Systems Engineering, Infotainment, led at Jaguar Land Rover. When Chris came on board, only three build engineers had the ability to release to production and it took four to six weeks to complete each release. Chris gradually scaled a CI/CD platform, step-by-step over two years, until every team across the organization was able to complete a release in less than 30 minutes. Today, the Jaguar Land Rover completes 50 to 70 deployments per day. This iterative pattern is seen over and over again with enterprise digital transformation. It’s encouraging because it means successful digital transformation can be driven by either executives top-down or engineers bottom-up.

    Keep the Fuel Burning

    According to Argyle, “91 percent of companies have at least started on digital transformation, but only half have created an organization-wide strategy.” This seems to align with Forrester’s metrics on half of the organizations that stall.

    So, how can organizations keep going and avoid the stall out?

    If I can make a confession, I am also one of those New Year’s resolution makers who just started working out in January. How it is that I’m still hitting the gym while others have dropped off? I have found that taking a “DevOps” approach is a key to my success. Instead of broad sweeping changes, I make small incremental improvements one day at a time.

    It turns out the strategy for getting started—“just commit”—is the same strategy to keep going.

    By starting small with a single project and gradually expanding scope in iterative steps, an organization can continue to improve over time at a rapid pace. Small, single-commit code changes coupled with CD increase throughput and stability. In the same way, small, incremental process improvements bring about the greatest success with digital transformation.

    — William Chia