# MetricReducer

Reduces metric volume by aggregating per-object time series into per-cohort summaries while preserving the original signal when anomaly rules fire. Series that don't belong to a monitored object flow through unchanged; series matching a drop rule are discarded.

**Type**: `object`

## Properties

- name: anomalies (optional)
  type: array of `AnomalyConfig`
  description: Anomaly rules.

- name: anomalyActions (optional)
  type: `AnomalyActions`

- name: cohorts (optional)
  type: `CohortsConfig`

- name: dropRules (optional)
  type: array of `DropRule`
  description: Drop rules — matching series are discarded outright.

- name: emitInterval (required)
  type: string
  description: How often the reducer flushes cohort-level output.

- name: hostMetadata (optional)
  type: `HostMetadataReporting`

- name: lateArrivalWaitDuration (required)
  type: string
  description: Watermark delay applied to give late-arriving data a chance to be aggregated.

- name: name (required)
  type: string

- name: objects (required)
  type: `ObjectsConfig`

- name: spaceAggOverrides (optional)
  type: object
  description: Per-metric-name overrides for space (cross-object) aggregation.

- name: timeGranularity (required)
  type: `TimeGranularity`

- name: type (required)
  type: enum: `metric-reducer`
  description: Metric reducer.