Grafana Dashboards

Grafana provides rich visualization and monitoring capabilities for the Nokia BNG Lab telemetry data.

Overview

Grafana is an open-source analytics and visualization platform:

Create interactive dashboards
Query Prometheus datasource
Set up alerts and notifications
Share dashboards with teams
Support for multiple datasources

Lab Configuration:

Container: grafana
Management IP: 10.77.1.14
Web UI Port: 3030 (mapped from 3000)
Version: 10.3.5
Default Credentials: admin/admin

Accessing Grafana

Web Interface

http://localhost:3030

Anonymous access is enabled - you can view dashboards without logging in. To edit dashboards or access admin features, log in with admin/admin.

Authentication Settings

Configured in lab.yml:

env:
  GF_ORG_ROLE: "Editor"              # Anonymous users are Editors
  GF_AUTH_ANONYMOUS_ENABLED: "true"  # Enable anonymous access
  GF_AUTH_ANONYMOUS: "true"
  GF_SECURITY_ADMIN_PASSWORD: admin  # Admin password

Pre-configured Dashboards

The lab includes two production-ready dashboards:

Dashboard Provisioning

apiVersion: 1

providers:
  - name: telemetry
    type: file
    updateIntervalSeconds: 10
    allowUiUpdates: false           # Dashboards are read-only
    options:
      path: /var/lib/grafana/dashboards
      foldersFromFilesStructure: true

Dashboards are automatically loaded from configs/grafana/dashboards/ at startup.

Note: allowUiUpdates: false means dashboard changes in the UI are not persisted. To modify dashboards permanently, edit the JSON files and restart Grafana.

1. SROS Dashboard

File: configs/grafana/dashboards/SROS-Dashboard.json
Devices: BNG1, BNG2 (Nokia SROS routers)

System Status
Port Statistics
BNG Sessions
Routing Protocols

CPU and Memory Monitoring

CPU utilization per device
Memory pool usage
System health indicators
Temperature and fan speeds

Panels display real-time resource consumption with color-coded thresholds.

2. SR Linux Dashboard

File: configs/grafana/dashboards/srlinux-telemetry-lite.json
Devices: Switch, OLT, TX (Nokia SR Linux devices)

Platform Metrics
Interface Statistics
Network Instances

Hardware Resource Usage

Control plane CPU per slot
Memory consumption
Application status
System processes

Datasource Configuration

Prometheus is pre-configured as the default datasource:

apiVersion: 1

datasources:
  - name: Prometheus
    type: prometheus
    url: http://prometheus:9090
    uid: prometheus
    isDefault: true

  - name: Loki
    type: loki
    url: http://loki:3100
    uid: loki

Verify Datasource

Navigate to Configuration (gear icon) → Data Sources
Select Prometheus
Scroll to bottom and click Save & Test
Should show: “Data source is working”

Loki is configured but not deployed in the lab by default. It can be used for log aggregation if added to the topology.

Navigating Dashboards

Dashboard Layout

Each dashboard consists of:

Variables: Dropdown filters (device, port, etc.)
Rows: Collapsible sections grouping related panels
Panels: Individual visualizations (graphs, gauges, tables)
Time Range: Selector in top-right corner

Using Variables

Dashboards include template variables for filtering:

Device: [ bng1 ▼ ]    Port: [ All ▼ ]    Interval: [ 5m ▼ ]

Variables update all panels when changed, allowing you to:

Focus on a specific device
Filter by interface
Adjust aggregation interval

Time Range Selection

Click the time picker (top-right) to select:

Quick ranges: Last 5m, 15m, 1h, 6h, 24h, 7d, 30d
Custom range: Absolute start and end times
Refresh interval: Auto-refresh every 5s, 10s, 30s, etc.

Example: “Last 1 hour” with ”10s refresh”

Zoom and Pan

Zoom in: Click and drag on a graph
Zoom out: Double-click graph
Pan: Shift + click and drag
Reset zoom: Click “Reset zoom” button

Creating Custom Dashboards

Due to allowUiUpdates: false, custom dashboards created in the UI are temporary. For permanent dashboards, export JSON and add to configs/grafana/dashboards/.

Create New Dashboard

Click + (plus) → Dashboard → Add new panel
Select Prometheus as datasource
Enter PromQL query
Configure visualization options
Save panel and dashboard

Example: CPU Usage Panel

Query
Visualization
Panel Settings

# Query builder or code mode
system_cpu{device="bng1"}

Or with aggregation:

avg by (device) (system_cpu)

Panel Type: Time seriesOptions:

Legend: {{device}}
Y-axis: Percent (0-100)
Thresholds: Yellow at 70%, Red at 85%
Fill opacity: 20%

Title: CPU UtilizationDescription: Real-time CPU usage per deviceRepeat options: Repeat by variable $device (optional)

Example: Interface Traffic Panel

# Bits per second
rate(port_statistics_in_octets{device="$device"}[5m]) * 8

Visualization: Graph with:

Ingress (positive) above X-axis
Egress (negative) below X-axis
Legend: {{port_id}} In/Out

Example: Session Count Gauge

sum(subscriber_mgmt_local_user_db_ipoe_session_stats_current)

Visualization: Gauge

Min: 0
Max: 1000 (or expected max sessions)
Thresholds: Green (0-500), Yellow (500-800), Red (800-1000)

Dashboard Variables

Create dynamic dashboards with variables:

Add Variable

Dashboard Settings (gear icon) → Variables → Add variable
Configure variable:

Name: device
Type: Query
Datasource: Prometheus
Query:

label_values(system_cpu, device)

Use in panels: {device="$device"}

Variable Examples

label_values(system_cpu, device)
# Returns: bng1, bng2, switch, olt

Dashboard Best Practices

Organize with Rows

Group related panels into collapsible rows:

System Health: CPU, memory, temperature
Interface Statistics: Traffic, errors, state
Protocol Status: BGP, ISIS, OSPF
BNG Metrics: Sessions, RADIUS, subscribers

Rows improve dashboard navigation and load performance.

Use Consistent Time Ranges

Ensure all panels use the same time range:

Don’t override panel time ranges unless necessary
Use dashboard-level time picker
Consider adding “Last updated” timestamp panel

Set Appropriate Refresh Rates

Balance freshness vs. load:

5s refresh: Real-time monitoring, troubleshooting
30s refresh: Active monitoring
1m+ refresh: Historical analysis
Off: Manual refresh for exploration

Add Context with Annotations

Display events on graphs:

Deployments
Configuration changes
Incidents
Maintenance windows

Requires annotation datasource (not included in lab).

Leverage Variables for Reusability

Create one dashboard for all devices:

Use $device variable in all queries
Add “All” option for overview panels
Chain variables: $device → $port

Exporting Dashboards

To make custom dashboards permanent:

1. Export from UI

Open dashboard
Click Share icon (top-right) → Export
Select Save to file
Download JSON file

2. Add to Lab

# Copy to dashboards directory
cp ~/Downloads/my-dashboard.json \
  configs/grafana/dashboards/my-dashboard.json

# Restart Grafana to load dashboard
sudo docker restart grafana

3. Version Control

# Commit to git
cd nokia-bng-lab
git add configs/grafana/dashboards/my-dashboard.json
git commit -m "Add custom BNG monitoring dashboard"

Alerting

Grafana supports alerting on metric thresholds:

Create Alert

Edit panel → Alert tab
Create alert rule
Define conditions:

WHEN avg() OF query(A, 5m, now) IS ABOVE 80

Configure notification channels (email, Slack, webhook)

Example Alert: High CPU

Condition:

avg(system_cpu{device="bng1"}) > 85

For: 5 minutes
Message: “BNG1 CPU utilization above 85% for 5 minutes”

Alerting requires notification channels to be configured in Grafana. This is not pre-configured in the lab.

Troubleshooting

Dashboard shows 'No Data'

Causes:

Prometheus datasource not connected
No metrics matching query
Time range has no data
PromQL syntax error

Solutions:

Test datasource connection
Verify query in Prometheus UI first
Check time range includes recent data
Review panel query for errors (red underline)

Dashboards not loading after changes

Problem: Modified JSON files not reflected in GrafanaSolutions:

Restart Grafana container:
```
sudo docker restart grafana
```
Check file permissions:
```
ls -la configs/grafana/dashboards/
```

Verify provisioning config:

sudo docker exec grafana \
  cat /etc/grafana/provisioning/dashboards/dashboards.yaml

Variables not populating

Problem: Variable dropdown shows “No options”Solutions:

Ensure query returns values in Prometheus
Check datasource is selected for variable
Verify label name matches metric (case-sensitive)
Test query in Explore page first

Slow dashboard loading

Problem: Dashboard takes long time to renderSolutions:

Reduce time range (e.g., 1h instead of 7d)
Decrease number of panels per row
Simplify PromQL queries (remove regex, use specific labels)
Increase scrape/sample interval
Use recording rules in Prometheus for expensive queries

Dashboard Examples

Subscriber Overview Dashboard

Purpose: Monitor BNG subscriber sessions Panels:

Total Active Sessions (Stat)

sum(subscriber_mgmt_local_user_db_ipoe_session_stats_current)

Sessions per BNG (Bar gauge)

sum by (device) (subscriber_mgmt_local_user_db_ipoe_session_stats_current)

Session History (Time series)

subscriber_mgmt_local_user_db_ipoe_session_stats_current{device=~"$device"}

Network Health Dashboard

Purpose: Overall lab health monitoring Rows:

Device Status: CPU, memory, uptime
Interface Status: Operational state, error rates
Protocol Health: BGP neighbors, route counts

Capacity Planning Dashboard

Purpose: Identify resource constraints Panels:

CPU trend over 7 days
Memory growth rate
Interface utilization percentiles (p50, p95, p99)
Session growth forecast

Next Steps

Available Metrics

Browse all metrics for dashboard queries

Dashboard Guide

Detailed guide to using pre-built dashboards

Stack Overview

Components

Metrics

Documentation Index

​Grafana Dashboards

​Overview

​Accessing Grafana

​Web Interface

​Authentication Settings

​Pre-configured Dashboards

​Dashboard Provisioning

​1. SROS Dashboard

​2. SR Linux Dashboard

​Datasource Configuration

​Verify Datasource

​Navigating Dashboards

​Dashboard Layout

​Using Variables

​Time Range Selection

​Zoom and Pan

​Creating Custom Dashboards

​Create New Dashboard

​Example: CPU Usage Panel

​Example: Interface Traffic Panel

​Example: Session Count Gauge

​Dashboard Variables

​Add Variable

​Variable Examples

​Dashboard Best Practices

​Exporting Dashboards

​1. Export from UI

​2. Add to Lab

​3. Version Control

​Alerting

​Create Alert

​Example Alert: High CPU

​Troubleshooting

​Dashboard Examples

​Subscriber Overview Dashboard

​Network Health Dashboard

​Capacity Planning Dashboard

​Next Steps

Available Metrics

Dashboard Guide

Build docs developers (and LLMs) love

Grafana Dashboards

Overview

Accessing Grafana

Web Interface

Authentication Settings

Pre-configured Dashboards

Dashboard Provisioning

1. SROS Dashboard

2. SR Linux Dashboard

Datasource Configuration

Verify Datasource

Navigating Dashboards

Dashboard Layout

Using Variables

Time Range Selection

Zoom and Pan

Creating Custom Dashboards

Create New Dashboard

Example: CPU Usage Panel

Example: Interface Traffic Panel

Example: Session Count Gauge

Dashboard Variables

Add Variable

Variable Examples

Dashboard Best Practices

Exporting Dashboards

1. Export from UI

2. Add to Lab

3. Version Control

Alerting

Create Alert

Example Alert: High CPU

Troubleshooting

Dashboard Examples

Subscriber Overview Dashboard

Network Health Dashboard

Capacity Planning Dashboard

Next Steps