Every year Puppet and DORA team up with a range of sponsors to assess the impact IT automation is having across a vast array of sectors and services. The annual “State of DevOps” report does this in terms of adoption and the effect that adoption is having. The report is based on a survey of over 27,000 IT professionals globally. The research categorizes the organisations the professionals work for into high, medium and low performers. They look at what distinguishes the performances of those organisations and draw some conclusions about potential root causes.
The key messages put forward in the report are:
The report shows an apparent narrowing in the field between the high and low performing IT departments. Last year the high performers released 200 times more frequently than the low performers. Now that difference is only a factor of 46. Similarly the gap in lead time to change has closed significantly. But we need to look at these factors alongside the quality statistics. The high performers are seeing a factor of 96 improvement in recovery time and a five-fold reduction in change failure.
It’s also interesting to see the less effective IT organisations are deploying monthly – six times more frequently than they were last year. That shows that many IT organisations are embracing DevOps. It’s how they go about implementing automation, and in particular what they choose to automate, not the decision to adopt DevOps which set the front runners apart.
Speed versus quality
These statistics show that all IT organisations are getting quicker at implementing automation, but the gap between the high-performing IT organisations and those at the lower end is widening in terms of quality. I believe the challenge on a DevOps journey, and the root cause of these statistics, is not the automation itself, but choosing what to automate and dealing with the implications of that automation.
When I work with clients I make sure they show me their problems, they don’t just tell me about them. If they say they have a problem they want to fix, I work with them to make sure they measure the extent and the impact of that problem. A client in the logistics sector told me they had a problem with slow deployments – they said they were taking three days. But this was their perception. In reality when we measured it, deployments were only taking half an hour, but they had to try 30 times before they got a successful one. They needed to focus on automating quality assurance in the stages before deployment – not automating deployment itself.
Leadership that leads to success
The principal message of the report is about leadership – because managing the impact of automation on an organisation can be harder than implementing it. Transformational leadership as a concept was introduced in 2004 by Alannah E. Rafferty & Mark A. Griffin from the School of Management at Queensland University of Technology, and suggests there are five dimensions to successful leadership: vision, inspirational communication, intellectual stimulation, supportive rather than instructive leadership, and personal recognition.
The reports says that delivering these five dimensions is a positive predictor for both commercial and non-commercial organisational success. This chimes with our own experiences dealing with both successful, and less successful leaders in the UK retail banking sector. Just thinking of meetings we had in two contrasting organisations: in the less successful, the leadership were very often those doing all the talking (ironically often speaking about empowering teams). The more successful bank had leaders who listened, asked questions, and offered advice.
Get on the high performers list
If you want to move your organisation up in the league of IT performance, you may need to rethink your leadership style – and consider carefully what you automate.