Originally reported on November 12, 2026.
LinkedIn has built an OpenTelemetry-based platform that connects user actions in its iOS, Android and web apps to the backend services handling them. The aim is to close a gap between client analytics, which can show a click or an error, and distributed traces, which show what happened inside services.
A page can generate many requests, each with its own OpenTelemetry traceId. LinkedIn adds a pageForestId to group requests from the same page context. Engineers can use the trace ID to inspect an individual request or the page-level ID to examine related activity affecting the user experience.
The platform uses OpenTelemetry across clients and backend services written in Java, Python, Go and Rust. LinkedIn wraps and extends standard SDKs to give product teams consistent instrumentation while handling mobile-app lifecycles, telemetry upload and LinkedIn-specific metadata. HTTP and gRPC frameworks instrument common request paths automatically; teams can add spans for important business operations. Telemetry travels through Apache Kafka and is stored in ClickHouse, where pageForestId, traceId and timestamps support correlation.
Because collecting every span would generate substantial data, LinkedIn uses coordinated head sampling to keep sampling decisions consistent across client and server activity. An anonymized user identifier helps maintain that consistency across a user journey. LinkedIn estimates that as much as 90% of some backend trace data could be classified as debugging-level data, rather than higher-value experience-quality telemetry.
On mobile devices, the platform batches spans and can flush telemetry when an app enters the background, before it may be suspended or terminated. It also uses attribute allowlists, blocklists and redaction to limit sensitive information entering traces.
The approach builds on capabilities found in existing tracing and application-monitoring tools. Its distinguishing design choice is to use OpenTelemetry as the common foundation for client and backend telemetry, with pageForestId adding a grouping layer above individual traces.
Source: infoq.cn (originally published 2026-11-12 16:44 UTC+8)


