Design telemetry approach for Juno

[stranger80]

Create a design for telemetry capability in Juno notes, including:

• Consider libraries and utilities
• Ensure "opt-in"is available for Juno users
• Take into consideration the future p2p network monitoring capability

[stranger80]

1. Use OpenTelemetry architecture (https://opentelemetry.io/docs/what-is-opentelemetry/)

   • Design "signals" for Juno
     • Traces
       • all RPC calls???
       • P2P operations???
         • P2P message handler execution?
       • ...consider (estimate) the volume of telemetry data!!!
     • Metrics
       • ...list needed...
     • ...skip Logs

2. Consider "passive" vs "active"

   • "passive" - Juno exposes OT endpoints and can be instrumented by an external OT Collector agent (eg. Prometheus)
   • "active" - Juno actively uploads telemetry data to a designated collector endpoint (eg. see https://github.com/open-telemetry/opentelemetry-go/tree/main/example/otel-collector)

3. Implement OT instrumentation framework in Juno

   • https://opentelemetry.io/docs/instrumentation/go/

4. Implement 1st iteration of traces and metrics

5. Measure the performance impact

   • Need to test the node RPC performance with and without telemetry to quantify the infrastructure increase required to host RPC SaaS

6. Design the OT Collector infrastructure

   • Collector agent or
   • Collector backend instance

7. Implement the OT Collector infrastructure

8. Implement Grafana integration for telemetry data

9. Develop Grafana dashboards

[stranger80]

Also take into account the recent discussion on "quick telemetry" as discussed in Brussels: https://github.com/NethermindEth/starknet-node-data-spec

[github-actions[bot]]

This issue is stale because it has been open 35 days with no activity. Remove stale label or comment or this will be closed in 14 days.