Skip to content

awsec2LNS.md

Latest commit

 

History

History
70 lines (52 loc) · 3.62 KB

awsec2LNS.md

File metadata and controls

70 lines (52 loc) · 3.62 KB
Raw

ChirpStack & LoRaWAN Local Deployment Upgrade Plan

This document summarizes the current cloud infrastructure for the ChirpStack & LoRaWAN setup and proposes an on‑premises upgrade to support 500 nodes and 50 gateways.

1. Current Cloud Infrastructure

Name Type vCPU RAM Purpose
lora small t3a.small 2 2 GB Test/Staging server – used for testing new ChirpStack configs and updates before moving to production.
lora large 1 t3a.large 2 8 GB ChirpStack Network Server (NS) – manages LoRaWAN MAC layer, ADR, device sessions, and downlink scheduling.
lora large 2 t3a.large 2 8 GB ChirpStack Application Server (AS) – decrypts payloads, provides APIs, and publishes to MQTT.
lora medium 1 t3a.medium 2 4 GB PostgreSQL Database – stores device/session data, join info, and ChirpStack metadata.
lora medium 2 t3a.medium 2 4 GB EMQX MQTT Broker – handles uplink/downlink messaging between ChirpStack components and external applications.
lora medium 3 t3a.medium 2 4 GB Redis/Memcached – caches session data and accelerates ChirpStack performance.

Current load:

  • Nodes: ~150
  • Gateways: ~30
  • Traffic: 1 packet (100 bytes) every 15 s → ~576 kB per node per day
  • Storage needs: ~25–30 GB/month (including logs and indexes)

2. Target Load After Upgrade

  • Nodes: 500
  • Gateways: 50
  • Traffic: 1 packet (100 bytes) every 10 s → 8,640 packets/day per node

Estimated storage:

  • Raw payloads: ≈ 13 GB/month
  • With logs, indexes, and metadata: ≈ 120–140 GB/month
  • 6‑month retention: ≈ 700–800 GB

3. Proposed Local Infrastructure (Upgraded)

Component vCPU RAM Storage Notes
ChirpStack Network Server 8 16 GB 100 GB Handles MAC layer, ADR, and sessions
ChirpStack Application Server 8 16 GB 100 GB Decrypts payloads, API, MQTT output
PostgreSQL Database 8 16 GB 600 GB Stores device/session data & metadata
EMQX MQTT Broker 4 8 GB 100 GB Handles high-volume MQTT messaging
Redis/Memcached 4 8 GB 50 GB Caching for performance
Testing/Staging Server 2 4 GB 50 GB Optional development environment

Total: 32 vCPUs, 64 GB RAM, ~1 TB SSD storage.

4. Key Improvements

  • Capacity Increase: From 150 → 500 nodes (over 3× scaling).
  • Consolidation: Fewer, more powerful servers replacing 6 smaller EC2 instances.
  • Performance: More vCPUs and RAM for ChirpStack, PostgreSQL, and MQTT under higher load.
  • Resilience: Ability to separate database and MQTT broker for fault isolation.
  • Future‑proofing: Scalable storage and compute to support growth beyond 500 nodes.

5. Recommendations

  • Use NVMe SSDs for PostgreSQL for better write performance.
  • Separate database and MQTT broker onto dedicated hardware or VMs for reliability.
  • Enable daily database backups and weekly off‑site replication.
  • Use TLS for MQTT and enforce strict firewall rules.
  • Integrate Prometheus/Grafana for monitoring system performance.