IoT Platform | VectorScaleDB

Protocol Adapters

Native support for IoT protocols. No middleware translation layer, no schema mapping — publish sensor data and VectorScaleDB handles the rest.

MQTT

JSON and CBOR payloads

Subscribe to MQTT topics and automatically ingest sensor telemetry. Supports both JSON and CBOR binary payloads for bandwidth-constrained environments. Entity IDs are extracted from topic hierarchy or payload fields.

Automatic entity ID extraction from MQTT topics
JSON and CBOR payload decoding
Configurable field-to-dimension mapping
QoS-aware acknowledgment

Industrial Protocols

OPC UA, Modbus, CoAP

Built-in adapters for industrial automation protocols. Connect directly to PLCs, SCADA systems, and industrial gateways without an intermediate broker. Each adapter normalizes readings into temporal vectors with consistent dimensionality.

OPC UA subscription with node browsing
Modbus TCP/RTU register polling
CoAP observe for constrained devices
Prometheus metric scraping

IoT-Optimized Compression

IoT data is uniquely compressible. Most sensors report stable readings most of the time. VectorScaleDB exploits this with drift-based compression that understands sensor behavior.

Idle Sensors

1,000x+ compression

A temperature sensor reporting 25.1°C every second for 8 hours generates 28,800 readings. VectorScaleDB stores this as a single compressed segment. Only genuine state changes create new segments.

Active Sensors

50-100x during drift

When a sensor enters a changing regime — a motor spinning up, a valve opening — compression adapts automatically. Drift-based segmentation captures the transition faithfully while still achieving 50-100x over raw storage.

Intelligent Write Elimination

Zero writes for stable sensors

Proprietary write elimination automatically detects when incoming readings represent no meaningful change. Stable sensors generate zero storage I/O during steady state, dramatically reducing infrastructure costs for large IoT deployments.

Fleet Analytics

Five built-in query functions designed for IoT fleet management. No ETL pipelines, no data warehouses — analytics run directly on compressed sensor data.

Anomaly Detection

Sensor anomaly scoring

Automatically detect sensors behaving outside their historical baseline. Anomaly scores are computed from compression drift metrics, so detection improves as more data flows in. No training step required.

POST /v1/iot/sensor-anomaly
Baseline computed from historical compressed segments
Severity scoring with configurable thresholds

Predictive Maintenance

Failure prediction from behavioral drift

Track how sensor behavior evolves over time. Rising drift rates often precede equipment failure. VectorScaleDB correlates behavioral changes across sensor fleets to identify maintenance needs before failures occur.

POST /v1/iot/predictive-maintenance
Drift trend analysis across device populations
Remaining useful life estimation

Fleet Health

Dashboard-ready fleet overview

Get a single-call summary of fleet health: device counts by status, top anomalies, compression statistics, and behavioral distribution across the entire device population.

POST /v1/iot/fleet-health
Aggregated health scores across all devices
Top-N anomalous devices with context

Correlation & Geofence

Cross-sensor analysis

Discover correlations between sensors using elastic temporal alignment and statistical correlation analysis. Run geofence queries to find devices within spatial boundaries and analyze their behavioral patterns.

POST /v1/iot/sensor-correlation
POST /v1/iot/geofence
Elastic temporal alignment for time-shifted signals

Edge Deployment

A single binary with no external dependencies. Deploy VectorScaleDB at the edge alongside your devices — no JVM, no Python runtime, no database server to manage.

Single Binary

Zero dependencies

Compiled Rust binary with embedded storage. No PostgreSQL, no Redis, no Kafka. Copy the binary to an edge gateway and start ingesting. Runs on ARM64 and x86_64.

Resource Efficient

Minimal footprint

Compression happens at the edge, reducing bandwidth to the cloud by 200–16,500x. The compression engine auto-sizes to available memory — runs on 256 MB devices or 256 GB servers.

Docker Profiles

Pre-built edge containers

Docker Compose profiles for single-node edge deployment. Health checks, graceful shutdown, and automatic recovery on restart. Production-ready out of the box.

Real-Time Subscriptions

Push notifications when sensor behavior changes. Subscribe to entities or entity types and receive events as they happen — no polling required.

WebSocket

Per-entity and per-type subscriptions

Open a WebSocket connection and subscribe to specific devices or entire device classes. Receive push events for new vectors, regime changes, and anomaly detections in real time.

Subscribe by entity ID or entity type
Filter events by severity or event type
Multiple subscriptions per connection

Reconnection

Replay buffer for catch-up

If a client disconnects and reconnects, VectorScaleDB replays missed events from a server-side buffer. No data loss during network interruptions — critical for remote IoT deployments with intermittent connectivity.