Time Series Aggregation

Time series aggregation (TSA) aggregates metric data at ingestion, before storage. Matching rules can apply to metrics emitted by instrumentation as well as metrics generated by signal-to-metrics.

TSA always runs in Dash0, not in SignalControl Edge. Metrics sent by an edge collector are aggregated after they reach Dash0, so TSA does not reduce network egress from your environment.

Aggregation reduces metric volume in two ways:

• Spatial aggregation: Remove selected attributes and merge series that differ only by those attributes, reducing cardinality.
• Temporal aggregation: Combine data points within configured time intervals. Using an interval longer than the input reporting interval reduces the frequency of stored data points.

The metric type determines how values are combined. For example, gauges keep the most recent value rather than summing values across sources. Summary metrics are not aggregated and pass through unchanged. See Aggregate Time Series for type-specific behavior and rule configuration.

Removing an attribute means you can no longer group or filter the stored metric by it. Keep the dimensions you need for your queries.

Aggregation also introduces delay before data is available to queries. Availability depends on the aggregation interval and configured delay, not a fixed latency guarantee. Emitting metrics directly from your application does not bypass a matching TSA rule.

Further Reading

• Aggregate Time Series for configuring aggregation rules.
• Common Metric Query Issues for handling delayed metric data in queries.
• Signal-to-Metrics for generating stored metrics from logs and spans.