Blog
Practical OpenTelemetry and observability engineering
OTLP support is not OpenTelemetry support
Accepting OTLP only means a vendor opened a port and can decode the wire format. Being OpenTelemetry-native means preserving and understanding the data model: semantic conventions, a shared resource model, and version drift.
Weaver from zero to hero: a getting started guide
Adopt OpenTelemetry Weaver to turn hand-maintained telemetry constants into a versioned schema with generated code and CI-enforced contracts.
Six OpenTelemetry talks worth watching from KubeCon EU 2026
Our picks from KubeCon EU 2026 in Amsterdam: three applied observability talks showing real-world OTel deployments, and three core project updates that affect every Collector operator.
Route your logs by severity with the OpenTelemetry Collector
How to use the OpenTelemetry Collector's routing connector to split logs by severity, sending critical logs to your vendor and archiving the rest to cheaper storage.
You don't have too much telemetry. You have bad telemetry.
Why observability cost problems are governance problems, not volume problems. A practical framework for identifying and eliminating telemetry waste before reaching for sampling.