Skip to main content
Version: 3.0 (next)

Three-layer namespace

MaestroHub runs at every layer: an edge per line, a hub per site and an enterprise hub for the group. Each layer adds its part of the namespace (line, then site and area, then enterprise) and buffers what it sends upward.

Scenario
5 of 5
Shape
Tier 1 per line · Tier 2 per plant · Tier 3 for the group
Applies to
3.0

Choose it when you want cross-plant KPIs on one namespace while every line and every site keeps working on its own.

Three-layer namespaceTier 1 per line · Tier 2 per site · Tier 3 for the group
  • MaestroHub
  • Reads machines
  • Data
  • Buffered on disk
  • Management
  • Your tooling
L5EnterpriseL4BusinessL3.5Plant DMZL3OperationsL2SupervisoryL1ControlBursaStore-and-forward: queued on the sender's disk while the far side is downStore-and-forward: queued on the sender's disk while the far side is downStore-and-forward: queued on the sender's disk while the far side is downStore-and-forward: queued on the sender's disk while the far side is downStore-and-forward: queued on the sender's disk while the far side is downFleet ManagerOptional · every deploymentLicences · healthCloud data platformSnowflake · Databricks · OneLakeMaestroHub · Tier 3Enterprise hub · every site under one rootCross-site model · KPIs · appsKubernetes · 3 nodes · data centreMaestroHub · Tier 2Site hub · Bursa DMZSite model · MES and ERP feedsTimescaleDB · EMQX · PrometheusK3s · one nodeMaestroHub · Tier 2Site hub · IzmirSite modelMaestroHub · Tier 1Press line edgeLine UNS · queueNative serviceMaestroHub · Tier 1Paint line edgeLine UNS · queueNative serviceS7 PLCsPressKeyencePressOPC UA robotsPaintEtherNet/IPPaintMQTTNATS or KafkaSnowflakeheartbeat

What you install

LevelComponentHow it runsRecommended hardware
L4Enterprise hub, Tier 3Helm chart with production values on three nodes3 nodes, each 16 vCPU, 24 GB RAM, 500 GB SSD
L3.5Site hub, Tier 2, one per plantThe single-node chart on K3s16 vCPU, 24 GB RAM, 500 GB SSD
L3Edge, Tier 1, one per lineOne binary as a native service8 vCPU, 16 GB RAM, 200 GB SSD
L5Fleet Manager (optional)Its own binary or Helm chartA small server or container host

Edges publish to their site hub over MQTT. Site hubs forward to the enterprise hub over NATS or Kafka. Every hop is opened by the lower layer, and every write that fails waits in a Store & Forward buffer on the sender.

When something fails

What failsWhat happens
The WAN between a site and the groupThe site keeps working and buffers its feed. Cross-plant KPIs lag until the backlog is delivered.
A site hubIts edges keep their lines running and buffer toward it. Tier 2 has one node; use Tier 3 at sites that cannot wait.

Limits. Three layers to operate and upgrade; keep their versions in step. Agree the topic contract between the layers (who owns which part of the namespace) before the first site goes live.

Previous: Independent plants · Next: Deployment scenarios · How to read the drawings: Deployment scenarios