Sign in Get started

DevOps Tutorial · DevOps Basics · Lesson 2 of 25

DevOps Culture and DORA Metrics

The principles behind DevOps and the four DORA metrics teams use to measure delivery performance.

Create a free account to save this page, take notes and track your progress. Sign in

The principles

A common summary is CALMS:

  • Culture: shared ownership of production; blameless learning from incidents.
  • Automation: builds, tests, deployments and infrastructure are automated.
  • Lean: small batches, limited work in progress, fast feedback.
  • Measurement: decisions are based on data from the pipeline and production.
  • Sharing: knowledge, tools and on-call are shared between developers and operators.

The DORA metrics

The DORA research programme found four metrics that predict software delivery performance:

Metric Question Elite teams
Deployment frequency How often do we deploy to production? On demand, many times a day
Lead time for changes Commit to running in production? Less than a day
Change failure rate What share of deployments cause a failure? Low (around 5% or less)
Time to restore service How fast do we recover from a failure? Less than an hour

Speed and stability go together: teams that deploy often in small batches also fail less and recover faster. Measure teams to improve the system, never to rank people, or the numbers get gamed.

DORA metrics from a deployment logTry it yourself
from datetime import datetime
from statistics import median

deploys = [  # (commit time, deployed time, caused an incident?)
    ("2026-03-02 09:10", "2026-03-02 11:00", False),
    ("2026-03-02 13:30", "2026-03-02 14:05", False),
    ("2026-03-03 10:00", "2026-03-03 16:40", True),
    ("2026-03-04 08:45", "2026-03-04 09:30", False),
    ("2026-03-05 15:20", "2026-03-05 15:55", False),
]
fmt = "%Y-%m-%d %H:%M"
lead_hours = [(datetime.strptime(d, fmt) - datetime.strptime(c, fmt)).seconds / 3600 for c, d, _ in deploys]
days = 5
print(f"deployment frequency: {len(deploys) / days:.1f} per day")
print(f"median lead time:     {median(lead_hours):.1f} hours")
print(f"change failure rate:  {sum(f for *_, f in deploys) / len(deploys):.0%}")

Output

deployment frequency: 1.0 per day
median lead time:     0.8 hours
change failure rate:  20%

Test yourself

Which is NOT one of the four DORA metrics?

  1. Deployment frequency
  2. Lead time for changes
  3. Lines of code written per developer
  4. Change failure rate
Show answer

Answer: Lines of code written per developer

DORA measures delivery outcomes: frequency, lead time, change failure rate and time to restore.

Was this page helpful?