Central Services

While edge devices handle execution at the point of operation, IronFlock’s central services provide the fleet-wide capabilities that require persistent infrastructure, a global view, or coordination across devices.

Central services run either in the IronFlock managed cloud or on your own infrastructure (Kubernetes or Docker Compose) for on-premises deployments.

Core Services

FleetDB — Time-Series Data Cluster

FleetDB is a cluster of PostgreSQL databases with TimescaleDB extension that stores all telemetry data collected from your fleet of devices. The cluster architecture enables horizontal scaling — each project gets its own database instance, and additional database nodes can be added as the fleet grows. It provides:

• Per-project database isolation — Each project gets its own dedicated database instance with separate tables and credentials. Data from one project cannot leak into another, enforced at the infrastructure level.
• Automatic provisioning — When you create a project and install an app that produces data, FleetDB automatically creates the necessary database, tables, schemas, and access credentials. No manual database setup required.
• Time-series optimization — TimescaleDB hypertables handle high-frequency sensor data with automatic partitioning, compression, and retention policies.
• Horizontal scaling — The cluster distributes database instances across nodes, so growing your fleet doesn’t bottleneck on a single database server.
• Scheduled transformations — Configure aggregations and data transformations that run on a schedule, producing derived tables from raw telemetry.

FleetDB Service — Data Processing & Dashboard Backend

The FleetDB Service sits between FleetDB and the rest of the system:

• Ingests data streams from devices via the WAMP message broker and writes them to FleetDB
• Evaluates alarm rules on incoming data in real time, triggering notifications when conditions are met
• Serves dashboard queries — The web UI’s charts, tables, and gauges fetch their data through this service
• Handles data transformations — Runs scheduled aggregations and derived calculations

Backend Service — Fleet Management

The backend manages all fleet-level operations:

• Device registration, grouping, and lifecycle management
• Project and user account management
• App releases, versioning, and OTA deployment orchestration
• App Store and marketplace operations
• Access control and privilege enforcement
• Audit logging

AI Service — Dynamic Agent Orchestration

The AI service powers IronFlock’s conversational AI capabilities. It orchestrates AI-agent conversations, dynamically adding and removing specialized agents as the conversation evolves:

• Dynamic agent management — Agents are added to a conversation when their expertise is needed and removed when they are no longer relevant, keeping conversations focused and efficient
• Routes natural language queries to the appropriate specialized agent
• Invokes device-side functions via the message broker for physical AI — the AI can read sensors, send commands, and interact with edge applications in real time
• Generates live chart widgets directly in conversation
• Manages multi-agent conversations with shared context and permissions
• Connects to LLM providers (Anthropic Claude, OpenAI GPT, Google Gemini)

Web UI — Browser-Based Control Plane

The IronFlock web UI is a single-page application where operators:

• Monitor device status, connectivity, and health across the fleet
• View real-time dashboards with sub-second latency
• Manage apps, releases, and OTA updates
• Configure alarms and notification routing
• Develop apps using the built-in cloud IDE
• Interact with the AI assistant

Container Registry

A private Docker registry that stores all app images. When you create a release of an app, the built image is pushed to the registry. When a device receives an OTA update, it pulls the image from the registry.

How Central Services Connect

All central services communicate through the WAMP message broker. This is the same broker that edge devices connect to, which means:

• A dashboard query in the web UI triggers a request through the broker to the FleetDB Service
• An AI command to read a sensor on a device travels through the broker to the device agent
• A device publishing telemetry data sends it through the broker to the FleetDB Service for storage

There is one unified communication layer for the entire system — not separate APIs for cloud-to-cloud and cloud-to-edge.

Deployment Modes

See On-Premises Deployment for private infrastructure setups.