Culture is not something you can implement; it is a byproduct of an environment that encourages specific behaviours over others. Perhaps this is why the culture change required for DevOps is the hardest aspect of an organisations DevOps transformation. You can invest time and money and have the best tools in place, but without the right culture, these tools will not bring you many benefits.
Considering DevOps is all about removing silos, we chose to start our cultural transformation journey by eliminating the barriers that exist between teams. We first looked at where the silos were and observed how our cross-functional development teams and operations teams communicated during and after an outage. It was evident that during these moments, communication was the most crucial part of resolving an issue quickly. We managed to improve this using a simple outage channel by inviting representatives from each team (e.g., Product leaders and Senior Developers) to this channel. Whenever support or operations detected an outage, it was posted to this channel. This meant that all teams had visibility on issues that were affecting customers, making all teams aware of the increased collaboration between different members of the organisation. In this way, the outage was no longer one person’s responsibility; rather, a cross-functional team collaborated on finding the source of the outage and supported one another until it was fixed.
Post outage reports
A post outage (or postmortem) report is excellent way to increase collaboration within your organisation. It is an opportunity for teams to learn from mistakes in an environment that is open, honest and, most importantly, without blame.
We ran our first post outage report session with five different team members, including Product owners and Cloud and Web Architects. We started the learning session by leveraging the ‘five whys’ approach to dig deeper into the cause of the outage. This quickly helped us identify areas we needed to invest in. The outcome of the report was a clear action plan with JIRA tickets and due dates. The due dates ensure we do not just reflect on what happened but take action to ensure they do not happen again. This could mean pausing development work on that shiny new feature and spending time on addressing tech debt that may have been put off for a while.
To summarise, the path to DevOps culture has no defined road. You pave it as you go, relying on the open communication and collaboration throughout the organisation. When something goes wrong, your open and blameless culture will not only ensure that you fix the issue quickly, but it will also significantly reduce the chances of the same issue happening again.