OpenTelemetry's 'Metrics, Logs, and Traces' Model Causes Observability Confusion

Observability Confusion in Tech

There is significant confusion within the tech industry regarding observability concepts, such as the definitions and relationships between traces, spans, logs, and metrics. This confusion persists despite efforts by observability systems like Honeycomb to clarify these topics.

OpenTelemetry's Role in Confusion

The author attributes much of this confusion to OpenTelemetry's presentation and approach, specifically its initial and clear distinction between metrics, logs, and traces. While acknowledging OpenTelemetry as a valuable solution, the author contends that its framework makes observability appear overly complex.

OpenTelemetry defines itself as a collection of APIs, SDKs, and tools for instrumenting, generating, collecting, and exporting telemetry data (metrics, logs, and traces) to analyze software performance and behavior.

Over-Complication of Traces

OpenTelemetry's detailed explanation of traces, often accompanied by complex diagrams, leads users to view traces as distinct and requiring significant learning. This focus on specific trace terminology, including spans, root spans, and nested spans, contributes to a perception of complexity, evidenced by OpenTelemetry's glossary containing over 60 terms.

Limitations for Distributed Systems

The author questions whether the 'logs, metrics, and traces' focus adequately represents the full power of observability, especially for distributed systems at scale. For such systems, the ability to 'slice and dice' data is more critical than rigid categorization.