Find costly Azure Data Factory activity patterns
On this page
Choose a factory, pipeline, or activity for design review by comparing activity-run counts, execution duration, copy DIU-hours, and data-flow vCore-hours for one defined period. Treat the comparison as directional unless the selected runs fully cover the matching factory scope and period.
Why this is worth a look
Azure Data Factory documents separate charging bases for orchestration activity runs, copy activities on Azure Integration Runtime using DIUs and execution duration, and data-flow execution and debugging using compute type, vCores, and execution duration. Cost Analysis is the factory-level cost view. Data Factory Monitor reports consumption for individual pipeline runs, while activity output can expose billableDuration. Together, these views help distinguish orchestration volume from movement or transformation consumption, without treating Monitor consumption as the charged price.
Run this check
CHECKLISTA read-only worksheet for Azure Cost Analysis and Data Factory Monitor. Record meter consumption and cost for one defined period, then use fully covered runs for reconciliation or partial coverage for prioritization.
Azure Data Factory cost pattern review
Execution surface and permissions
[ ] Run this worksheet manually in Azure Cost Analysis and the Azure Data Factory Monitor experience. It performs viewing and recording only.
[ ] Confirm at least read access to Azure cost data and access to the data factory Monitor experience.
Assumptions and scope
[ ] Select one Cost Analysis scope and one complete reporting period. Record the period, factory scope, and displayed time zone.
[ ] Use the same period and time zone in Monitor. Treat the period and scope as the comparison key.
Cost Analysis
[ ] Review Azure Data Factory costs at the selected Cost Analysis scope. Use available grouping fields to keep the comparison scope consistent.
[ ] Record actual cost separately from usage quantity and unit. Keep activity runs, DIU-hours, and vCore-hours as separate measures.
[ ] Note the ADF meter families present, including orchestration activity runs, copy DIU-hours, and data-flow vCore-hours.
Pipeline and activity review
[ ] In the data factory Monitor experience, select the same period and time zone.
[ ] For selected material pipeline runs, open the consumption report. Record each ADF meter and its consumed amount.
[ ] Compare pipeline name, status, start and end times, duration, and rerun history. Record repeated executions separately.
[ ] Open a selected pipeline run, open its activity runs, and inspect JSON output for billableDuration where present.
[ ] For copy activity, record DIU usage and execution duration. For data flow activity, record compute type, vCore count, and execution duration when available.
Decision record
[ ] If the selected runs cover all relevant runs for the same factory, period, and time zone, compare totals with Cost Analysis. If coverage is partial, use run data only to prioritize a design review and do not label differences as unassigned usage.
[ ] Name the pattern as orchestration volume, long execution, copy DIU consumption, or data-flow compute consumption.
[ ] Record the owner and reliability requirement before proposing a schedule, batching, retry, or runtime change.How to confirm it
- 01
Set scope, units, and period
Use one Cost Analysis scope and one complete reporting period. Review Azure Data Factory costs at that scope and use available grouping fields to keep the comparison consistent. Record cost separately from usage quantity and unit, including activity runs, DIU-hours, and vCore-hours. You need at least read access to view Azure cost data.
- 02
Review pipeline consumption
Open the Azure Data Factory Monitor experience, select the same period and time zone, and review the material pipeline runs. Open the consumption report for each selected run and record the amount consumed for each ADF meter. Compare pipeline name, status, start and end times, duration, and rerun history.
- 03
Inspect activity evidence
Open a selected pipeline run to view its activity runs, then open an activity's JSON output and look for billableDuration where present. For copy activities, record DIU usage and execution duration. For data flow activities, record compute type, vCore count, and execution duration when available.
- 04
Reconcile only complete coverage
If the selected runs cover all relevant runs for the same factory, period, and time zone, compare their totals with Cost Analysis. Otherwise, use the run data only to prioritize a design review and do not treat differences as unassigned usage. Focus on orchestration volume, long execution, copy DIU consumption, or data-flow compute consumption, and record the owner and reliability requirement before proposing a change.
Before making changes
Assume you can read both Cost Analysis and Data Factory Monitor, and that the reporting period, factory scope, and displayed time zone are recorded consistently. Monitor shows meter consumption for a pipeline run, not its actual charged price, which depends on Azure account type and currency. Integration Runtime charges are prorated by minute and rounded up, so raw execution duration may not map directly to billed duration or cost. Cost Analysis can include scope beyond the selected runs, so reconcile only with complete matching coverage.