Micro Focus is now part of OpenText. Learn more >

You are here

You are here

4 steps to successfully starting a DevOps transformation

John Jeremiah Evangelist, GitLab
4 steps to make your DevOps project a success

Embarking on an organizational transformation with DevOps can be a massive undertaking.

In fact, it can be downright overwhelming when you consider the scope and size of the changes it requires. DevOps represents a major cultural change, so keep your expectations in check. Don't expect the organization to change overnight, and don't expect the entire organization to change at the same time. Consider following these four steps for a smooth transition.

1. Start at the right place at the right time

Decide where you'll plant the seeds of change and grow your future DevOps leaders and experts. Where can you harvest quick wins and learn what works and what doesn't? Start with the business:

  • Does it demand speed and velocity from IT?
  • Is management desperate to go faster?
  • Are they open to change?

If not, you may never get off the ground. Next, look at the IT team supporting the business. Are they receptive or resistant to change? Will they adopt changes and apply DevOps discipline to the way they work? This is a prerequisite.

Then, consider the technology the IT teams are using. If possible, try to choose areas where technology is more cloud and web oriented. It's not that DevOps is only suitable for web and cloud; it's that newer technology with lower legacy debt will make it easier for your first time out the gate. Finally, look for an area where the teams will be empowered. This area should allow them to drive change, own the results, and know when to get out of the way and trust their teams.

Now, to take a page from the shampoo commercials, it's time to "lather, rinse, and repeat," following the wisdom of USAF Col John Boyd's Observe Orient, Decide, Act (OODA) loop.

2. Lather (observe and orient)

Observe and measure how things work today. How long does it take for a new requirement to get to production? Find the areas with the greatest pain, and find the biggest bottlenecks. Get some data, create a baseline, and figure out how to update it. One word of caution here: Don't get stuck measuring for the sake of measurement. Do just enough, and then get moving.

3. Rinse (decide and act)

Do something different. The team knows where the problems are, so fix those. If the problem is in coding and builds, work on source control and continuous integration. If testing and QA are the biggest issues, automate and implement continuous testing to streamline those processes. If the problem area is the deployment and delivery of apps and infrastructure, continuous delivery is the place to start.

As you start making changes, don't forget to pay attention to the data. If the data gets better, then keep going. If things get worse, don't hesitate to go back to the drawing board.

4. Repeat

Don't rush things. Give the process enough time to sink in with the team. It may even be a month or more before you can go back to the beginning. Any team needs to use the new process and tools several times before they'll feel confident that it works better than it did before. When your team is finally ready, start again with the data and search for new bottlenecks and pain points. This is your next opportunity to improve, accelerate, and streamline delivery.

Your participation in the DevOps transformation will vary with your role in the organization. IT leaders need to empower the team so they can own the change and the results. The partnership between IT leadership and the team must be built on mutual trust based on a shared goal of delivering more value to the customer.

But don't stop there. The improvement loop for DevOps has no end: It's really all about continuous improvement.

Keep learning

Read more articles about: Enterprise ITIT Ops