Ep. #93, Adaptive Sampling Strategies with Mike Goldsmith
On episode 93 of O11ycast, Mike Goldsmith talks to Martin Thwaites and Ken Rimple discuss Honeycomb's contribution to the OpenTelemetry Collector, the Adaptive Tail Sampling Processor. This processor was created based on years of experience building Honeycomb's custom, industry leading tail sampler, Refinery.
By: Ken Rimple
Join other developers and engineering leaders who are sharing what’s actually working in the agent era. Come to Honeycomb Observability Day in London, on October 6-7, 2026. Attend workshops on day 1, and a full track of practitioner talks on day 2.
Details
On episode 93 of O11ycast, Mike Goldsmith talks to Martin Thwaites and Ken Rimple about Honeycomb's contribution to the OpenTelemetry Collector, the Adaptive Tail Sampling processor. This processor was created based on years of experience building Honeycomb's custom, industry leading tail sampler, Refinery.
Mike Goldsmith is an engineer at Honeycomb who has worked in and around OpenTelemetry since its early days. His work has spanned OpenTelemetry SDKs and specifications, with a recent focus on improving the OpenTelemetry Collector and developing tools that help teams operate telemetry pipelines more efficiently and effectively.
Show notes are available on Heavybit.
Links
- Bringing the Most Advanced Sampling to the OpenTelemetry Collector (Honeycomb blog post by Mike Goldsmith)
- Honeycomb
- OpenTelemetry
- OpenTelemetry Collector
- OpenTelemetry Collector Contrib
- OpenTelemetry Drain Processor
- OpenTelemetry Log Deduplication Processor
- OpenTelemetry Signal to Metrics Connector
- OpenTelemetry Dynamic Sampling Processor
- TraceState - Probability Sampling
- Can you afford to sample your production data? (Reddit)
- Honeycomb Refinery
- Honeycomb Refinery Sampling Methods
- CNCF Slack