September 2022 Summaries

The text discusses how to use DORA metrics to measure, learn from, and improve software development failures. It emphasizes the importance of collaboration among team members, including developers, designers, project managers, management, and ops, in addressing software failure. The article presents three steps for learning from software failure: measuring it, analyzing it, and making changes based on the findings. DORA metrics are introduced as a tool to measure engineering efficiency, with four key metrics: Deployment Frequency, Change Lead Time, Time to Restore Service, and Change Failure Rate. The text also explores different levels of software failure measurements, from incident management systems to more tactical metrics like error rates in Sentry or CPU usage in Datadog or Prometheus. It highlights the importance of a blameless postmortem process for learning from failures and making improvements. Finally, it encourages automation as a way to reduce mistakes and increase efficiency in software development processes.

DORA metrics are four key performance indicators that help assess a team's efficiency in software development. These metrics include Deployment Frequency, Change Lead Time, Time to Restore Service (Mean Time to Recovery), and Change Failure Rate. By measuring these metrics, teams can identify areas for improvement, track progress on initiatives, measure the impact of unrelated changes, and assess the effects of team scale events. However, it's crucial to remember that metrics should not be the answer but rather a tool to ask better questions and make real change in a team.

The DORA metrics originated from a question and hypothesis among a group of people who formed the DevOps Research and Assessment (DORA) startup to investigate best practices in software engineering teams. They published their findings in the first State of Devops Report in 2014, identifying four key metrics that correlate with high-performing teams: Deployment Frequency, Change Lead Time, Change Failure Rate, and Mean Time to Recovery. These metrics are now widely used as industry standards for measuring engineering efficiency. However, it's crucial to remember that DORA metrics aren't the only ones to consider; they should be tailored according to specific organizational needs and problems.

Cycle Time is an essential metric for improving productivity within a team, though not one of the four DORA metrics. It measures the time taken from a request being made to code being healthy in production. This broader and more inclusive metric than Change Lead Time addresses the needs of stakeholders who can request features and end users who expect software to work reliably. Cycle Time is crucial for improving productivity as it helps keep track of backlog size, ensures that the team delivers what the business wants, and allows for strategic planning and goal setting. By measuring and managing Cycle Time effectively, teams can improve their delivery speed and turn around features faster.