By Huseyin Dursun, VP of Engineering, and Shane Van Bentum, Director, Cloud Operations, both VMware
Since the turn of the century, the never-ending search for higher productivity and better efficiency brought about the movements of agility and DevOps—and traditional, siloed IT organizations have been looking for ways to embrace and adapt these approaches to transform their operations.
Agile practices and DevOps offer significant benefits if they are adapted correctly, but the transition is not always smooth for IT organizations, and VMware is no exception.
The way we were
Traditionally, VMware IT teams were loosely divided based on larger-scale themes. They focused on delivering monolithic-style services and projects with little-to-no interactions between groups. Ultimately, this approach proved untenable for long-term sustainability in the cloud era, where factors such as time, quality and agility are of paramount importance. This led to the adoption of the DevOps approach to stay competitive.
However, there were significant shifts involved in the transition, and not all of them technological.
What we had in mind
A successful DevOps transition requires changes to people, processes and tools. The people—and shift in the mindset of team members, leadership and culture—are often the hardest yet the best place to begin.
We started with people, and kept people in the center of our transition efforts. But resistance was naturally inevitable as with any enterprise paradigm change.
Like it or not, people are often the greatest barrier to adoption of DevOps
The resistance to change shows its face in a variety of ways, starting from the moment transition is announced. The syndrome of “this is how we have been doing it” is one of the most common. Another one is the rejection to be on-call to own reliability, resiliency and overall quality of service. Like all other major transformational changes, organizations need to be prepared to face higher-than-typical levels of attrition in their teams, something that is unfortunately an unavoidable consequence.
This doesn’t mean VMware made wholesale changes overnight, with no input from those affected. Team leadership redefined roles, processes and expectations in subtle ways, with an underlying goal of truly empowering every employee at every level, even to the point of supporting failure as a way of learning. We intentionally stayed away from textbook definitions in our adaption process, but used Edward Deming’s “14 Points for Management” as guiding principle, especially the very last point:
- Put everybody in the company to work to accomplish the transformation.
Making it a reality
We started with our terminology to ensure that we are always speaking the same language. The result was a nomenclature for the new culture and working model, something we call VMware IT’s view of DevOps.
We focused these efforts on three main areas:
- Build-and-Operate where engineers own the services they create, and then either reap the rewards of a well-built service or suffer the pain of cutting corners. This means all engineers must share the burden of an on-call rotation.
- Elimination of toil where everything is automated that can be automated, or feasible to automate.
- Make self-healing a priority so that when issues happen, engineers must troubleshoot, runbook them (human automation), script them to be event-driven, and then implement code fires to remediate it.
What did we learn on this journey?
We underestimated the time and complexity involved with asking our operations people to incorporate automation. Writing code wasn’t the challenge that we had feared—the challenges revolved around the processes, additional tooling and thinking like a developer. We also learned that uniform coding and documentation standards, dividing our work into sprints, and creating meaningful success criteria are equally necessary to be successful.
The benefits of change
Embracing the culture DevOps—and the shift to Infrastructure as Code—has had many benefits, the primary one being that we are now able to maintain consistency across services. This increases reliability and reduces production incidents. Runbooks are turning into code, and self-healing is becoming ingrained into our services.
We are truly on a journey. With the advantages becoming visible, our cultural transformation accelerating, and no one wants to go back anymore.
VMware on VMware blogs are written by IT subject matter experts sharing stories about IT’s transformation journey using VMware products and services in a global production environment. Visit our portal to learn more.