Signal-to-metrics rules convert matching logs and spans into metric data points at ingestion, counted and aggregated before storage. You can derive metrics from telemetry you already emit without adding metric instrumentation to your application. Events reported as log records can match log rules; events are not a separate input signal type.

When conversion happens

Rules apply to incoming telemetry once they take effect. Creating a rule does not backfill metrics from spans or logs already stored in Dash0. Conversion runs in Dash0 or, with SignalControl Edge, inside your own network before telemetry leaves it.

This differs from synthetic metrics, which Dash0 computes at query time from retained telemetry. Signal-to-metrics produces stored metrics that you query with PromQL.

When to use signal-to-metrics

Count matching log records or measure span duration without changing application code.
Record numeric values from span or log attributes as sums, gauges, or distributions.
Keep aggregated characteristics of spans or logs after the raw telemetry expires, when your metric retention is longer.

The generated metrics follow the target dataset's metric retention, not the retention of the source logs or spans. See Datasets for retention settings.

Dimensions and timing

Choose which resource and signal attributes to keep as metric dimensions. Each distinct combination produces a time series, so retain only attributes you need for grouping or filtering. You cannot recover discarded dimensions from the stored metric.

Aggregation and export introduce delay before the metric is queryable. Account for the configured aggregation interval and observed ingestion delay when choosing alert windows. Time series aggregation can add further delay if it applies to the generated metric. See Common Metric Query Issues for query guidance.

Signal-to-Metrics

When conversion happens

When to use signal-to-metrics

Dimensions and timing

Further Reading