Can Pipeline analytics isolate step-level queue delays from aggregate job duration?

Can Azure DevOps Pipeline analytics isolate step-level queue delays?

Standard Azure DevOps Pipeline analytics aggregates timing data at the job level. The dashboard displays total pipeline duration and overall queue time, but it does not break down queue delays per individual step within a job. If your team relies on the UI to attribute wait times, step-level queue contention may be misinterpreted as logic‑related slowdown.

Confirmed facts

• The built‑in analytics view shows queue time as a single metric per completed job.
• Step‑start, step‑end, and queueTime fields are not surfaced in the Pipeline analytics UI.
• Misattributing agent contention to step logic is possible when only job‑level duration is visible.

Supported path for step‑level telemetry

1. Retrieve each step’s startTimestamp, endTimestamp, and queueTime via a common Azure DevOps REST API endpoint (shape may vary by instance configuration).

GET /_apis/pipelines/runs/{runId}/jobs/{jobId}/steps?api-version=6.0-preview.1

Examine the queueTime field for every step. The sum of per‑step queue times may differ from the job’s total queue time, especially when parallel steps run on separate agents.

Verification step

1. Compare the aggregate of step‑level queue times with the job’s total queue time shown in analytics. A significant gap often indicates distributed agent wait that the UI hides.

Missing diagnostic detail

Does the pipeline job contain steps that run on the same agent or across multiple agents? This determines whether step‑level queue time reliably reflects agent contention or gets diluted across parallel execution paths. In single‑step pipelines, no internal queue delay exists to measure.