EDA Telemetry Lab

The EDA Telemetry Lab demonstrates how to leverage full 100% YANG telemetry support integrated with EDA (Event Driven Automation). In this lab, Nokia SR Linux nodes are dynamically configured via EDA and integrated into a modern telemetry and logging stack that includes Prometheus, Grafana, Promtail, Loki, Alloy and Kafka exporters for alarms and deviations.

eda-telemetry-hero.mp4

• EDA-Driven Configuration: Automate SR Linux configuration and telemetry subscriptions with EDA.
• Modern Telemetry Stack: Export telemetry data using EDA’s Prometheus exporter and monitor alarms/deviations via the Kafka exporter.
• Enhanced Logging: Capture and aggregate system logs using Promtail, Alloy and Loki.
• Deployment Options: Deploy with either Containerlab (clab) for live traffic or CX (Simulation Platform) for license-flexible testing.
• Traffic: Generate and control iperf3 traffic to see dynamic network metrics in action.

Lab Components

• Fabric: A Clos topology of Nokia SR Linux nodes.
• Telemetry: SR Linux nodes stream full YANG telemetry data. EDA exports these metrics via its Prometheus exporter and sends alarms/deviations using its Kafka exporter.
• Visualization: Grafana dashboards (with the Flow Plugin) provide real-time insights into network metrics.
• Alarms: Data is collected by Kafka, processed by Alloy and aggregated in Loki, with logs viewable in Grafana.
• Traffic Generation: Use iperf3 tests and provided scripts to simulate live traffic across the network.

Deployment Variants

There are two variants for deploying the lab:

• Using Containerlab: The simulated SR Linux nodes, client containers and the telemetry stack are deployed using Containerlab. This deployment variant requires a license for EDA, as the simulated SR Linux nodes are deployed outside of EDA. On the positive side, using Containerlab allows for live traffic generation using iperf.
• Using CX (EDA Simulation Platform): The simulated SR Linux nodes are spawned in EDA CX and thus do not require a license. The telemetry stack for convenience is deployed using Containerlab. This deployment variant does not require a license but does not support traffic generation with iperf.

This README focuses on the Containerlab deployment variant as it leverages iperf-generated traffic for demonstration purposes. See cx/README.md for the CX deployment variant.

Important

EDA Installation Mode: This lab requires EDA to be installed in the Simulate=False mode. Ensure that your EDA deployment is configured accordingly. EDA Version: This lab is designed for EDA version 25.4. Ensure you are using this version or later. Hardware License: A valid hardware license for EDA version 25.4 is mandatory for using this lab with Containerlab.

1. Ensure kubectl is installed and configured: To test if kubectl is installed and configured, run:

   kubectl -n eda-system get engineconfig engine-config \
   -o jsonpath='{.status.run-status}{"\n"}'

   You should see Started in the output.

2. Initialize the Lab Configuration:

   Run the provided init.sh which does the following:

   • ensures uv and clab-connector tools are installed
   • retrieves the EDA IP and sets it in the configs/prometheus/prometheus.yml files.
   • save EDA API address in a .eda_api_address file.

   ./init.sh

3. Deploy containerlab topology:

   Run containerlab deploy to deploy the lab.

4. Install the EDA Apps (Prometheus and Kafka):

   Run:

   kubectl apply -f manifests/0000_apps.yaml

   TIP: Depending on your setup this can take couple of seconds/minutes. Please check in the EDA UI if the apps are installed.

5. Integrate Containerlab with EDA:

   Run:

   clab-connector integrate \
   --topology-data clab-eda-st/topology-data.json \
   --eda-url https://$(cat .eda_api_address)

   TIP: Check Clab Connector docs for more details on the clab-connector options.

6. Deploy the Manifests:

   Apply the manifests:

   kubectl apply -f manifests

7. Enjoy Your Lab!

Tip

Shutdown interfaces via WebUI: Client 1, exposes the port 8080 for the WebUI. You can use the WebUI to shutdown interfaces on the SR Linux nodes.

Accessing Network Elements

• SR Linux Nodes: Access these devices via SSH using the management IP addresses or hostnames (e.g., ssh clab-eda-st-leaf2).

• Linux Clients: Access client-emulating container via SSH: e.g. ssh user@clab-eda-st-server3 (password: multit00l).

Telemetry & Logging Stack

Telemetry

• SR Linux Telemetry: Nodes stream full YANG telemetry data.
• EDA Exporters:
  • Prometheus Exporter: EDA exports detailed telemetry metrics to Prometheus.
  • Kafka Exporter: Alarms and deviations are forwarded via EDA’s Kafka exporter, enabling proactive monitoring and alerting.
• Prometheus: Stores the telemetry data. The configuration (located in configs/prometheus/prometheus.yml) is updated during initialization.
• Grafana: Visualize metrics and dashboards at http://grafana:3000. For admin tasks, use admin/admin.

Logging

• Alloy & Loki: Alloy processes Kafka data and sends it to Loki for storage. Alloy has a web interface at http://alloy:12345.
• Promtail & Loki: Collect and aggregate SR Linux syslogs (e.g., BGP events, interface state changes). Logs are accessible and filterable via Grafana.
• Prometheus UI: Check out real-time graphs at http://prometheus:9090/graph.

Traffic Generation & Control

The lab includes a traffic script (named traffic.sh) that launches bidirectional iperf3 tests between designated clients.

Test Pairs Configured:

• Server4 to Server1:

  • From server4 (clab-eda-st-server4) targeting server1’s IP 10.10.10.1 on port 5201
  • From server4 targeting server1’s VLAN interface 10.20.1.1 on port 5202

• Server3 to Server2:

  • From server3 (clab-eda-st-server3) targeting server2’s IP 10.10.10.2 on port 5201
  • From server3 targeting server2’s VLAN interface 10.20.2.2 on port 5202

Default Test Settings:

• Duration: 10000 seconds
• Report Interval: 1 second
• Parallel Streams: 10
• Bandwidth: 120K
• MSS: 1400

Usage Examples:

• Start Traffic: To launch tests from a specific client or from both clients, run:

  ./traffic.sh start server3
  ./traffic.sh start server4
  ./traffic.sh start all

• Stop Traffic: To stop tests on a specific client or on all clients, run:

  ./traffic.sh stop server3
  ./traffic.sh stop server4
  ./traffic.sh stop all

EDA Configuration

The lab includes several manifest files that define the configuration of EDA apps and the network fabric:

• Apps Installation (0000_apps.yaml): Installs the Prometheus exporter (prom-exporter v2.0.0) and the Kafka exporter (kafka-exporter v2.0.1).
• Interfaces (0009_interfaces.yaml): Configure LAG interfaces on SR Linux leaf switches.
• TopoLinks (0010_topolinks.yaml): Define LAG topology links hanging off of leaf switches towards the servers.
• Prometheus Exporters (0020_prom_exporters.yaml): Prometheus Exporter configuration to stream telemetry metrics (CPU, memory, interface status, routes, etc.).
• Kafka Exporter (0021_kafka_exporter.yaml): Kafka Producer configuration to stream alarms and deviations.
• JSON-RPC server config (0025_json-rpc.yaml): Configlet with the JSON-RPC server configuration to enable port shutdown automation.
• Syslog (0026_syslog.yaml): Set up syslog forwarding to a centralized server.
• Fabric Topology (0030_fabric.yaml): The Clos-based fabric configuration with eBPG/iBGP underlay/overlay.
• Virtual Networks (0040_ipvrf2001.yaml and 0041_macvrf1001.yaml): Virtual networks configuration to support IP VRF and MAC VRF services.

Conclusion

The EDA Telemetry Lab offers a modern, automated approach to network telemetry and logging by integrating SR Linux with EDA. With fully automated configuration, a powerful monitoring stack leveraging EDA’s Prometheus and Kafka exporters, and flexible deployment options, this lab is an ideal starting point for exploring event-driven network automation.

Happy automating and exploring your network!

---

Connect with us on Discord for support and community discussions.