Work / NexusTrace
Case study 06
NexusTrace
Tracing and service health for architectures where failure has nowhere obvious to point.
The problem
When a request crosses six services, an error rate tells you that something broke. It never tells you where, and the logs of each service on their own never add up to the story.
The system
A Go backend carrying OpenTelemetry-compatible span models, with a Next.js and TypeScript dashboard over WebSockets. Traces render as a waterfall timeline, dependencies as an interactive service graph, and logs correlate to traces by ID so a failing request can be followed end to end. Alert rules fire on error rate, latency and health.
- Backend
- Go 1.22, OpenTelemetry-compatible spans
- Frontend
- Next.js 14, TypeScript, WebSocket live updates
- Storage
- PostgreSQL 16, optional
- Views
- Dashboard, traces, metrics, logs, dependencies, alerts
- Demo
- Built-in simulator generates realistic microservice traffic
The result
Requests per second, error rate and P95 and P99 latency stream live. A CLI covers every operation the API exposes, so the platform is usable without opening the dashboard at all.
Let's build something useful.
Send the problem, not a spec. I reply with an approach and a scope I can hold to.