Function Execution Metrics¶
Azure Functions publishes two execution-focused platform metrics for Consumption-plan apps: FunctionExecutionCount (how often functions ran) and FunctionExecutionUnits (a combined measure of execution time and memory). Together they explain both throughput and cost on the Consumption plan.
flowchart TD
A[Function Invocation] --> B[FunctionExecutionCount]
A --> C[FunctionExecutionUnits]
C --> D[Execution Time x Memory]
D --> E[GB-seconds Billing]
B --> F[Invocation Volume Trend]FunctionExecutionCount¶
The number of times functions in the app executed within the aggregation window.
| Property | Value |
|---|---|
| REST name | FunctionExecutionCount |
| Unit | Count |
| Aggregation | Total (Sum) |
| Time grain | PT1M (1 minute) |
| Scope | Function apps only |
Use the Sum aggregation to count invocations. This metric answers "how much work did the app do?" and is the primary signal for throughput trends and traffic anomalies.
FunctionExecutionUnits¶
A combination of execution time and memory consumption, expressed in MB-milliseconds. This is the metric that underpins Consumption-plan billing.
| Property | Value |
|---|---|
| REST name | FunctionExecutionUnits |
| Unit | Count (MB-milliseconds) |
| Aggregation | Total (Sum) |
| Time grain | PT1M (1 minute) |
| Scope | Function apps only |
Memory is not a separate metric
On the Consumption plan, memory consumption is not exposed as its own Azure Monitor metric. It is folded into FunctionExecutionUnits. To observe memory directly, query the Application Insights performanceCounters table (Private Bytes) — see Application Insights Telemetry.
Converting Execution Units to GB-seconds¶
Consumption billing is expressed in GB-seconds. FunctionExecutionUnits is reported in MB-milliseconds, so:
The divisor combines the MB-to-GB conversion (1,024) and the millisecond-to-second conversion (1,000). Averaged memory is rounded up to the nearest 128 MB for billing, so the metric is an observability signal rather than an exact invoice figure.
Plan Support Caveats¶
| Plan | FunctionExecutionCount | FunctionExecutionUnits |
|---|---|---|
| Consumption (Windows) | Supported | Supported |
| Consumption (Linux) | Not available | Not available |
| Flex Consumption | Use OnDemand*/AlwaysReady* metrics instead | Use OnDemand*/AlwaysReady* metrics instead |
| Premium / Dedicated (Linux) | Not supported | Not supported |
For Flex Consumption apps, the equivalent signals are split across always-ready and on-demand instances — see Flex Consumption Metrics.
Alerting Guidance¶
- Throughput drop: alert when
FunctionExecutionCount(Sum, PT5M) falls to zero during expected traffic hours — a common signal of a broken trigger or misconfigured binding. - Cost spike: chart
FunctionExecutionUnits(Sum) daily; a sustained increase without a matchingFunctionExecutionCountincrease indicates functions are running longer or using more memory per call.