InfluxDB + Grafana Stack for LoRaWAN

Why InfluxDB for LoRaWAN

LoRaWAN generates time-series data. Temperature reading at 10:00, another at 10:15, another at 10:30. Thousands of sensors, all timestamped, all the time.

You could store this in PostgreSQL or MySQL. People do. It works poorly. Those databases aren't designed for continuous streams of timestamped measurements. Queries get slow, tables get huge, performance degrades.

InfluxDB is built for time-series data. Handles millions of datapoints efficiently. Queries that take 30 seconds in PostgreSQL complete in under a second. It's the right tool for the job.

Standard stack I use: LoRaWAN Network Server (ChirpStack) → InfluxDB → Grafana.

Data Pipeline

From network server to InfluxDB:

• ChirpStack has InfluxDB integration built-in (config file setting)
• TTN needs middleware (Node-RED, custom script, or integration service)
• Helium requires API polling or webhook receiver
• Write sensor data as measurements with device_id tag

Example InfluxDB schema:

measurement: temperature
tags: device_id, location
fields: value (float), battery (int), rssi (int)
timestamp: nanosecond precision

Grafana Dashboard Design

Network health dashboard:

• Gateway uptime percentage (last 24h, 7d, 30d)
• Packet count per gateway (time series graph)
• Map panel showing gateway locations with status colors
• Alert list showing offline devices

Sensor data dashboard:

• Current temperature/humidity/whatever (stat panels)
• Historical trends (line graphs, 24h default, selectable range)
• Multi-axis graphs (temperature + humidity on same panel)
• Threshold annotations showing alarm levels

Device management dashboard:

• Table of all devices with last-seen timestamp
• Battery level gauge panels
• RSSI/SNR signal quality over time
• Data rate distribution (SF7 vs SF12 usage)

Data Retention

Don't keep raw 1-minute data forever - InfluxDB fills disk fast.

Retention policy strategy:

• Raw data: 30 days
• 5-minute averages: 1 year
• Hourly averages: 5 years

Use InfluxDB continuous queries or Telegraf aggregation to downsample automatically.

Common Mistakes

No indexing tags properly: Always tag device_id and location. Queries filtering by field values are slow.

Writing too frequently: Transmitting every 10 seconds fills database unnecessarily. Most applications work fine with 10-30 minute intervals.

Not monitoring InfluxDB itself: Use Telegraf to monitor InfluxDB CPU/memory/disk. Database going down means you lose all current data.

What I Provide

Services:

• InfluxDB installation and configuration
• Data pipeline setup from your network server
• Grafana dashboard design
• Query optimization
• Retention policy configuration
• Alert rule setup

You own everything:

• Self-hosted InfluxDB + Grafana
• Complete configuration documentation
• Dashboard JSON exports
• No monthly fees