Accelerate: DevOps Metrics & Delivery Science learning path: everything to study, topic by topic

The four DORA measures of software delivery performance: deployment frequency, lead time for changes, change failure rate, and time to restore service — balancing throughput and stability.

How often an organization successfully releases to production; a proxy for batch size and the team's ability to deliver in small, frequent increments.

The time from code committed to code running in production; measures the speed of the delivery pipeline from commit to deploy, distinct from product lead time.

The percentage of changes to production that result in degraded service and require remediation (rollback, hotfix, patch); a stability measure of quality, not raw defect counts.

How long it takes to recover from a failure in production; recognizing that failure is inevitable, this measures resilience and the ability to restore service quickly.

The operational availability/reliability measure added in later DORA research — meeting reliability targets (SLOs) — that pairs with the four delivery metrics to capture operational performance.

DORA cluster analysis groups organizations into elite/high/medium/low performers; elite performers outperform on all four metrics simultaneously, disproving the speed-vs-stability tradeoff.

Keeping application code, configuration, system configuration, and build/deploy scripts in version control — a foundational capability that predicts delivery performance.

Working in small batches on trunk or short-lived branches merged daily, avoiding long-lived feature branches; one of the strongest predictors of high delivery performance.

Reliable automated tests that developers trust and primarily create/maintain; tests run continuously to give fast feedback, with failures signaling real defects.

Integrating code frequently into a shared mainline with automated builds and tests on every commit, keeping the build green and surfacing integration problems early.

Automating the deployment process so releases are repeatable, push-button, and low-risk — reducing manual steps that cause errors and lengthen lead time.

Integrating security and infosec review early in the delivery process and into the work of development teams, rather than as a late-stage gate; predicts both security and delivery performance.

Architecture that lets teams test and deploy on demand without depending on other teams' systems or fine-grained coordination; testability and deployability are the key architectural outcomes.

Teams that can make large-scale changes to their systems' design, choose their own tools, and deploy without permission from outside the team — autonomy enabled by architecture.

Limiting work in progress to expose bottlenecks, reduce context switching, and improve flow; WIP limits are effective only when combined with visual feedback on flow.

Visual displays of quality, productivity, and work status combined with monitoring of system and business metrics, giving teams the information to make decisions and improve.

Peer review and pair programming outperform external change-approval boards (CABs); heavyweight approval processes correlate negatively with delivery performance and do not improve stability.

Working in small batches, gathering and implementing customer feedback, making flow of work visible, and empowering teams to change specs without approval — enabling experimentation.

Westrum's typology of pathological (power-oriented), bureaucratic (rule-oriented), and generative (performance-oriented) cultures; generative culture predicts software delivery and organizational performance.

Leadership characterized by vision, inspirational communication, intellectual stimulation, supportive leadership, and personal recognition; it enables (but cannot substitute for) the technical practices.

Continuous delivery and generative culture reduce burnout and deployment pain; identity, autonomy, and reducing toil improve wellbeing, which in turn sustains high performance.

How the DORA findings were validated: cross-sectional survey research, latent constructs, psychometrics, cluster analysis, and inferential/predictive analysis establishing that capabilities drive outcomes.

Using delivery metrics to learn and improve rather than to compare or punish teams; avoiding vanity metrics, gaming, and Goodhart's law by focusing on global outcomes and team-level improvement.