Architecture

System design

Separation without friction

Leafcuttr splits management, control, and data planes so edge workloads stay lean, operators stay in control, and data movement stays intentional.

Management
Console, Kubernetes Operator, API Fleet provisioning, lifecycle actions, and configuration changes happen here.
Coordinates the fleet
Control
Leafcuttr Controller + Leaf Monitor Bootstrap, change management, and metrics flow over HTTP and WebSocket.
Keeps the edge in sync
Data
Leafcuttr Server Kafka, MQTT, and HTTP traffic stays local first, then replicates asynchronously when connectivity allows.
Streams at the edge

Management Plane

The Console, Kubernetes Operator, and API coordinate fleet-level provisioning and lifecycle management without touching the data path.

Control Plane

The LeafCuttr Controller communicates with Leaf Monitor on each edge host over HTTP and WebSocket for bootstrap, configuration, change management, and metrics.

Data Plane

The LeafCuttr Server accepts Kafka, MQTT, and HTTP traffic at the edge, then replicates asynchronously to the mothership when connectivity is available.

1 Provision

Define the fleet and push the initial configuration.

2 Coordinate

Monitor edge hosts, adjust settings, and keep changes consistent.

3 Stream

Keep data local, then replicate upstream when the network cooperates.