aggregation-temporality · research · also Metric aggregation temporality, Metric temporality

Aggregation temporality

Aggregation temporality specifies whether a metric aggregate covers a new reporting interval or accumulates measurements from a fixed start time.

Related Observatory article: Read on hesela.com

OpenTelemetry uses delta and cumulative temporality for sums and histograms, including exponential histograms; gauges have no aggregation-temporality field. Delta points describe reporting intervals, while cumulative points retain a common start until accumulation state resets. [1]

Successive cumulative snapshots already share earlier observations. Adding their counts counts those observations again. For compatible snapshots from the same uninterrupted stream, subtraction can recover interval counts and sums, but not generally its minimum or maximum. [2]

Constructed example: a first report for [1, 100] ms has count 2 and sum 101 ms. Adding either [10, 20] or [14, 16] ms gives the same cumulative count 4, sum 131 ms, extrema 1/100 ms and bucket counts [3, 1] at boundary 50 ms. Subtraction gives count 2 and sum 30 ms, but the interval extrema differ. Adding both cumulative counts instead gives 6. [3]

Keep stream identity, units, start/end timestamps and bucket interpretation with each point. The conservative fixture rejects resets, incompatible layouts and overlapping or noncontiguous delta windows rather than inventing missing coverage; it is not a complete OTLP receiver. Detailed reset and overlap guidance in the reviewed data model remains Development. [2]

Relations

related
Exponential histogram (OpenTelemetry)

Citations

  1. OpenTelemetry Protocol metrics.proto: Sum, Histogram, ExponentialHistogram and AggregationTemporality; pinned d31f7acf. (accessed 2026-11-20)
  2. OpenTelemetry Metrics Data Model: Histogram, Temporality, Resets and Gaps, Overlap; pinned a0c5323b. (accessed 2026-11-20)
  3. Hesela: exact cumulative/delta histogram example and regression tests; no SDK execution. (accessed 2026-11-20)
Sources & caveats
  • Source synthesis and exact non-negative-integer examples. No exporter, collector, SDK or device measurement was executed.

JSON: /data/terms/aggregation-temporality.json

Published by Hesela team · Updated

Report a correction · Editorial standards · Reuse guidance

Cite this page
Hesela team. Aggregation temporality. Hesela. 2026-11-20. https://hesela.dev/terms/aggregation-temporality/