Measuring the right metrics from DevOps

DevOps gets a lot of lip service in IT circles, and rightly so. It’s the ideal environment for creating the innovative products, services and applications that will drive a competitive advantage.

But too often we fail to grasp the factors that make a transition to DevOps successful. Even if we do, then what defines success for a DevOps initiative; what helps us measure if we are improving? We measure our business performance, so why shouldn’t we measure our DevOps performance?

Matching metrics to your value proposition

Part of the solution will rely on DevOps organisations expanding their monitoring capabilities into previously unimagined areas. These can range from story point estimation to number of broken builds , and can even include the amount of overtime your DevOps are logging during a project. Because DevOps is designed to drive a competitive advantage, the ideal performance metrics will be those that best link up with your unique business model.

Value stream mapping is one approach which incorporates data on the performance of the tool chain, as well as traceability, to assess the DevOps lifecycle. With this data, senior leaders can measure the value of the process against KPIs such as failure mean-time and defect rate to create a visible measure of success.

Getting the right metrics in real-time

Equally important to using the right metrics is your ability to measure them in the right timeframe. Speed is one of the major influences in DevOps. The traditional cycle and response times of the past are now inadequate for today’s lightning fast digital economy, where disruption is the only constant.

DevOps organisations need to focus on building real-time feedback loops that integrate with automated systems for feeding into continuous delivery. We have the DevOps infrastructure already where everything can be code, including infrastructure, applications and APIs  and everything can be automated, including builds, tests and deployments (and I’m not saying that everything needs to be measured from the beginning either).  DevOps teams have the opportunity to monitor processes to identify areas of improvement, changing potential cycles from weeks to only hours.

Don’t overcommit to metrics

Remember that DevOps environments are agile by nature, so the definition of success needs to be relatively fluid. The metrics that underpin what the DevOps achieves enable them to improve by identifying the opportunities. Analytics and machine learning will also have a crucial role to play in automating iterative improvements during the development cycle.