Telemetry App Demo – Recogni App-Sim + OpenConfig (Telegraf) → OpenTelemetry Pipeline

This project is a proof-of-concept telemetry pipeline demonstrating how to collect fake hardware and application metrics and observe them end-to-end:

app-sim: a Rust service simulating Recogni app metrics (chip temperature, inference latency).

telegraf: scrapes simulated OpenConfig/Junos metrics (e.g., fan speed) via gNMI.

otel-collector: receives both OTLP metrics (from app-sim) and scraped metrics (from Telegraf), processes them, and exports to Prometheus.

prometheus: stores time series.

grafana: visualizes dashboards showing how chip temperature, fan speed, and latency relate.

The design makes it easy to swap fake sources (random jitter) with real telemetry when available.

Requirements

• Docker
• Docker Compose
• (Optional) Python 3 if running fake_fan.py locally for testing

Getting Started

1. Clone the repo git clone [email protected]:cooogus/telemetry-app-demo.git

2. Build and start services

docker compose up -d --build

3. Look at Grafana dashes and Prometheus stats through localhost ports 3000 and 9090