Measure deployment performance

Under Monitor, open Database Changes. All four figures follow the time range you select here, which defaults to 30 days.

Note: The same four figures appear in the Database Changes section of the Operations Dashboard, which has no time control. The three rate figures there cover the last 7 days and Cycle Time the last 30, so the two screens can disagree without either being wrong.

They sit in the tile row at the top of the dashboard.

Deployment Frequency and Cycle Time describe throughput. Change Failure Rate and Rollback Rate describe stability, and they move together when a problem is real: a failure rate climbing on its own more often means changes are failing before they land than after.

Cycle Time measures how long a database change takes to reach production. For each changeset, it is the elapsed time between the change's first successful deployment to a non-production environment and its first successful deployment to production. The metric shown on the card is the average across all qualifying changesets over the last 30 days.

A changeset is counted only when it has actually completed the journey to production:

• Only changesets that have reached production are included. A change that has been deployed to a non-production environment but not yet to production is still in flight and is excluded until it reaches production.

• Changes deployed straight to production are excluded. If a change reaches production without a prior non-production deployment, there is no non-production-to-production span to measure, so it does not contribute to the average.

• Rolled-back deployments are not counted; the first deployment that stays applied is the one that matters.

• Deployments through unclassified connections are excluded. If a database connection isn't assigned to an environment, its deployments can't be placed on the non-production → production journey, so they are not counted.

Production is determined by your existing environment classification. No extra setup is required. The supporting text under the metric shows how many of your recently production-deployed changesets had a measurable cycle time.

Note: Production is identified by the environment whose slug is prod, the default Production environment. Environment slugs are editable per workspace. If you rename or replace the prod environment, deployments there will no longer be recognized as production deployments by this metric.

Deployment Frequency, Change Failure Rate, and Rollback Rate each have a trend chart below the tile row. Selecting a tile switches the chart to that metric, and Deployment Frequency is selected when the page opens. Cycle Time has no chart.

Note: The chart always plots the last five weeks, whichever time range you have selected, so it and the metrics above it can describe different periods. Read the chart for the shape of the trend and the metric for the number.