The scenario that wastes the most engineering time goes something like this. A request is slow, or blocked, or returning the wrong response, and nobody knows exactly where in the stack it went wrong. Was it a security rule? A cache miss? A URL rewrite applied in the wrong order? A Worker that timed out? You end up jumping between logs, dashboards, and config screens, stitching together a picture that should have been one view from the start.
Cloudflare just addressed that problem directly with Cloudflare Traces, now in open beta. It extends automatic tracing across their entire platform, from the moment a request arrives at Cloudflare’s edge to the point it reaches your origin server. One timeline. No plugins, no custom instrumentation, no code changes required.
The key word is automatic. Distributed tracing, a technique where each step of a request’s journey through your infrastructure gets tagged with timing and outcome data, has historically required engineering work to set up. You annotate your code, configure agents, wire up a backend like Jaeger or Zipkin. Cloudflare generates all of this for you, covering security rule evaluations, cache decisions, URL transformations, Worker execution, and routing in a single request trace.
This matters for anyone running complex infrastructure on Cloudflare. Previously, Workers Tracing gave visibility into what happened inside Workers, but once the request left that scope you were back to guessing. Now you see the full picture. Security rule blocked a request? The trace shows which rule fired, how long the evaluation took, and the resulting action. Cache miss went to origin? The trace shows exactly where the 527 milliseconds went.
The implementation follows open standards. Cloudflare Traces uses OpenTelemetry, a widely adopted framework for capturing telemetry data across distributed systems, and supports W3C traceparent headers for end-to-end context propagation. That means you can follow a single request from your CDN through your backend services and into your database, with one consistent trace ID connecting the dots. You can export everything to any OTLP-compatible destination, so it integrates with Datadog, Grafana, Honeycomb, or whatever observability stack you already use.
Trace Rules let you control sampling. Set a base sampling rate for general traffic, then override it to capture 100% of requests matching specific criteria like a particular endpoint, IP address, or header value. Debugging a targeted issue becomes a precise operation instead of a fishing expedition.
The business case is simple. Faster debugging means shorter outages, which means less revenue lost and fewer customers who noticed something went wrong. The engineering teams that can pinpoint a problem in minutes rather than hours are not just more productive, they are actively protecting the bottom line every time something breaks in production.
Want to explore how better observability and DevOps practices could benefit your business? Let’s talk.