OpenTelemetry 1.25 vs Datadog 2026: Tracing Overhead for 1000 RPS Microservices Workloads

Distributed tracing is critical for debugging microservices, but instrumentation overhead can degrade production performance. This benchmark compares OpenTelemetry (OTel) 1.25 and Datadog’s 2026 tracing stack under a sustained 1000 requests per second (RPS) microservices workload to quantify real-world overhead.

Test Setup

We deployed a 3-service e-commerce microservices stack on Kubernetes (1.29):

Frontend: Node.js 20, handles user requests

Backend: Go 1.22, processes business logic

Database: PostgreSQL 16, persists data

Load was generated via k6 at a constant 1000 RPS for 60 minutes. We measured overhead with 100% trace sampling to isolate instrumentation impact, with no other observability tools running.…