wide-events

A wide event is a single canonical log line per request that consolidates identity, user, business, performance, and outcome context into one enriched event for distributed systems observability. This high-dimensional model makes cross-service correlation easy during incident debugging.

To design structured logging for incident debugging, attach high-dimensional queryable fields to trace spans covering identity, user, business, performance, and outcome attributes. This ensures emitted telemetry can answer incident-driven questions like which failures affected premium users under a specific feature rollout.

The best way to sample OpenTelemetry spans without losing critical signals is applying a tail sampling policy that retains errors, p99-slow requests, and VIP users while sampling the rest. This controls observability costs while preserving the telemetry needed for incident debugging.

Yes, you can use wide events to correlate logs across multiple microservices by consolidating fragmented, low-context logs into one comprehensive event per request. This high-dimensional model with structured span annotations replaces scattered logs optimized for writing with queryable events optimized for correlation.

Incident debugging takes long with fragmented observability logs because they are optimized for writing instead of querying, making it hard to correlate context across distributed services. Wide events solve this by consolidating request, user, business, performance, and outcome data into one enriched event.