Skip to main content

Managing High Cardinality

This document explains how Levitate provides visibility, superior defaults, and control levers to tame high cardinality

Levitate Limits

Quota TypeLimitReset PeriodPossible Actions
Per Time Series Cardinality1MPer HourCan be raised on request
Per Time Series Cardinality20MPer DayCan be raised on request
Streaming Aggregation Cardinality3MPer HourCan be raised on request
Ingestion Concurrency20KPer SecondCan be raised on Request
Number of Metrics Aggregated in one Pipeline1 MetricPer QueryCannot be changed for now
Time Series Scanned Per Query — Blaze Tier5MPer QueryCannot be changed
Time Series Scanned Per Query — Hot Tier10MPer QueryCannot be changed
Samples Scanned Per Query100MPer QueryCannot be changed
Query Time Range35 DaysPer QueryCan be raised on request

Even though Levitate can scale to large values of cardinality in terms of the ingest pipeline, large per-metric cardinalities have an adverse effect on read response-times.

Beyond a daily cardinality of 3M time series per metric, query response times for a metric start degrading.

It is advised to keep daily per-metric cardinality within this limit by using Levitate's streaming aggregation pipeline.

Read more about using PromQL powered Streaming Aggregations to reduce high cardinality metrics data.