Database Nodes

Database nodes receive data from the DDB topic structure and persist it into specialized storage backends. Each node acts as a consumer, subscribing to MQTT topics and writing data in a format optimized for its target database system.

Available Nodes

Node	Description	Compatible Payloads
Aveva PI	Industrial historian for OT environments — stores data in Aveva PI Asset Framework	historian (Sparkplug B)
Blob	Stores raw binary payloads into cloud object storage (S3, GCS, Azure Blob)	blob / any
Key-Value PostgreSQL	Persists key-value data into a relational PostgreSQL database	kv (JSON)
TimescaleDB	Time-series historian built on PostgreSQL/TimescaleDB for SQL analytics	historian (Sparkplug B)

Links

• Aveva PI Store Database Node
• Blob DWS (Data Warehouse Service)
• PostgreSQL Key-Value Store Database Node
• TimeScaleDB Database Node

How Database Nodes Work

All database nodes inherit from the BaseDatabaseNode class and share common behaviors:

1. Subscription — Each node subscribes to one or more MQTT topics based on its compatible payload type.

2. Decoding — Incoming messages are decoded (e.g., Sparkplug B payloads are parsed into metric key-value pairs).

3. Persistence — Decoded data is written to the target database using that system’s native API or SDK.

Choosing a Database Node

Need	Recommended Node
Long-term time-series analytics with SQL	TimescaleDB
Industrial OT historian (PI System)	Aveva PI
Simple relational storage for metadata / KV data	Key-Value PostgreSQL
Raw binary/object storage in the cloud	Blob (S3/GCS/Azure)

Related Links

• Payload Schema

• Architecture Overview

• mfi_ddb_database_nodes Source Code