Metrics Concepts
Understand metric fundamentals including temporality, cardinality, and how they affect pipeline design and downstream costs.
2 minute read
Overview
Metrics are numeric measurements that capture system behavior over time. While logs and traces provide detailed context, metrics enable efficient monitoring, alerting, and trend analysis with minimal storage overhead.
Edge Delta processes metrics from various sources (Prometheus, OTLP, application exporters) and transforms them before routing to destinations. Understanding two key concepts helps you design efficient pipelines and control costs:
- Temporality - How metric values relate to time (delta vs cumulative)
- Cardinality - The number of unique timeseries generated by a metric
Why these concepts matter
Cost implications
Most observability platforms bill by timeseries count, not metric name count. A single metric definition can generate thousands of timeseries depending on its attributes, directly impacting your bill.
Pipeline design
Different destinations have different requirements. Some reject cumulative metrics entirely; others expect specific aggregation patterns. Understanding temporality helps you configure the right processors.
Performance
High-cardinality metrics consume more memory, increase query latency, and can overwhelm downstream systems. Proactive cardinality management keeps pipelines healthy.
Metrics concepts
Learn the fundamentals:
- Metric Temporality - Understand delta vs cumulative metrics and aggregation rules
- Metric Cardinality - Understand how attributes multiply to create timeseries
Related resources
- Reduce Metric Cardinality - Strategies for controlling timeseries growth
- Cumulative to Delta Processor - Convert cumulative metrics to delta format
- Aggregate Metric Processor - Group and summarize metrics
- Rollup Metric Processor - Create single aggregated values