OpenTelemetry: Using Unified Semantics to Drive Insights + Project Update

Open Telemetry Integrations and Compatibility

• Open Telemetry has several streaming protocols and projects baked into the project itself
• Open Telemetry is interoperable with other projects and teams adding support for it
• Native integrations are starting to use native OTLP APIs
• Contrib repos have hundreds if not thousands of integrations with existing technologies

Structured logs with a regularized format make it feasible to generate metrics out of them, but the harder problem is parsing them as they may not be standard across different service types. The Elastic Common Schema is being worked on to get locked in and stable in the spec, and a validation and testing harness will be implemented to ensure everything conforms to the spec. Backward compatibility with Prometheus is a core goal of Open Telemetry, which is why slashes are not allowed in metric names.

This is the official OpenTelemetry session at Kubecon. OpenTelemetry is excellent at capturing traces, metrics, logs, and other signals at scale, however its capabilities go far beyond this. One often-overlooked piece of functionality is OpenTelemetry's semantic conventions, which consistently apply metadata to all signal types so that they can be correlated, aggregated, and more intelligently processed during Collector pre-processing and within observability analytics systems that receive OpenTelemetry data. In this session, we will discuss: - OpenTelemetry's core semantic conventions, what data is guaranteed, and how this data applies to different signal types (including logs) - How OpenTelemetry's semantics can be extended within an organization - Examples of the semantic conventions in action We will then present an update on OpenTelemetry's roadmap, what's new in the project, and host a maintainer Q&A.