ControlTheory co-founders Jon Reeve (Walter) and Eric Anderson (The Dude) presented at Open Observability Summit + OTel Community Day on June 26 in Denver.
Their talk, Taming Metric Cardinality: Practical Cost Reduction with the OpenTelemetry Collector, took a fun but technical deep dive into one of the most critical (and costly) challenges in observability: metric cardinality.
Why Metric Cardinality Matters
Metric cardinality refers to the number of unique time series generated by metrics. It often explodes due to high-cardinality tags like user_id, container_id, or request_id, creating massive cost and noise in your observability platform.
Whether you’re using Prometheus, Datadog, or another backend, high metric cardinality can:
- Drive up telemetry storage and processing costs
- Degrade query performance
- Obscure meaningful signals in dashboards and alerts
And it usually goes unnoticed—until the bill arrives.
Talk Overview: How to Tame Metric Cardinality
In this session, Jon and Eric walked through how to use the OpenTelemetry Collector and OTTL (OpenTelemetry Transformation Language) to take back control of your metrics pipeline:
- Identify high-cardinality metrics before they hurt you
- Use the filter processor to drop or modify noisy series
- Normalize attributes and remove sensitive or redundant tags
- Test changes using the OTTL Playground
- Implement safe rollout strategies using OpenTelemetry control planes
Get a copy of the slides here: https://static.sched.com/hosted_files/otelopenobservabilityna25/34/Taming_Metric_Cardinality.pdf
The result? Lower costs, clearer signal, and better observability hygiene—without changing a line of application code.
And yes, there will be The Big Lebowski references, and props!
Whether you’re already fighting metric cardinality explosions, or just starting your OpenTelemetry journey, this talk will give you practical tools to reduce cost and improve clarity in your telemetry pipeline.
Hope we saw you in Denver—or at the lanes after! Otherwise reach out and connect!
Back
press@controltheory.com
