Skip to main content
Version: 3.0 (next)

Deployment scenarios

Which MaestroHub servers to install, at which plant level, how many, and how they connect. Start with the layout closest to yours.

Server types
Tier 1 · Tier 2 · Tier 3 · Fleet Manager
Levels
Purdue L1 to L5, the plant DMZ as L3.5
Applies to
3.0

Five common layouts​

Each layout has its own page: the drawing, what to install, what happens when a part fails, and the limits.

Store-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 down
1 · Tier 1 · one deploymentLine-side edgeOne binary next to the line runs the whole plant.
Store-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 down
2 · Tier 1 per line · Tier 3 in the DMZEdge plus DMZ hubEdges on the lines, a highly available hub in the DMZ.
Store-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 down
3 · Tier 2 or 3, central · nothing installed at sitesOne hub, many sitesOne central instance; nothing installed at the sites.
4 · Tier 1 or 2 per plant · Fleet ManagerIndependent plantsEvery plant runs its own; Fleet Manager watches them all.
Store-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 down
5 · Tier 1 per line · Tier 2 per site · Tier 3 for the groupThree-layer namespaceMaestroHub at the line, the site and the group.

The servers​

Every layout is built from these four. A tier describes the shape of a deployment, not its location: any tier can run at a factory, in your data centre or in a cloud.

ServerWhat it isMachinesEach machine (recommended)Runs onHigh availabilityUsually sits at
Tier 1One process with every module inside, plus an embedded message bus, MQTT broker, historian and databases18 vCPU, 16 GB RAM, 200 GB SSDWindows, macOS or Linux, as a native service (Install and run)No. The service manager restarts it 5 seconds after a crash.L3, next to the line
Tier 2The same single process in one pod on a one-node K3s cluster, with TimescaleDB, EMQX and Prometheus beside it (deployment/context-platform/single-node). Manages many sites.116 vCPU, 24 GB RAM, 500 GB SSDLinux with K3sNoL3.5, the plant DMZ, or central (L4)
Tier 3One pod per module with 2 to 3 replicas, with clustered NATS, EMQX, TimescaleDB and Elasticsearch (deployment/context-platform/helm). Manages many sites.316 vCPU, 24 GB RAM, 500 GB SSDKubernetes 1.26 or laterYes, at the application tierL3.5, the plant DMZ, or central (L4)
Fleet ManagerOptional. One view of several separate deployments: health, licences and single sign-on settings (Architecture)18 vCPU, 8 GB RAM, 100 GBIts own binary or Helm chartNo. Servers keep their licence for 7 days without it.Central (L4 or L5), reachable over HTTPS

The minimums are on System requirements; measured rates for Tier 1 are on Sizing and performance. On Tier 3, high availability covers the application: pods, NATS, EMQX, TimescaleDB and Elasticsearch replicas. Spreading the nodes over racks, power and storage is part of your platform design.

Where each server goes and how they connect​

Every role on one drawing. Select a connection to light it up. Every connection between levels is opened by the lower one, so no inbound connection reaches the control network, and every write that fails waits in a Store & Forward buffer on the sender.

Where each server goesevery role at once
  • MaestroHub
  • Reads machines
  • Data
  • Buffered on disk
  • Management
  • Your tooling
L5EnterpriseL4BusinessL3.5Plant DMZL3OperationsL2SupervisoryL1ControlOne plantStore-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 · one per companyLicences · sign-in settings · healthCloud data platformsSnowflake · Databricks · Event HubsCentral hub · Tier 2 or 3one for the group · data centre or cloudEvery site in one deploymentERP · MES · LIMSbusiness systemsSite hub · Tier 2 or 3one per plant, in the DMZJoins the lines · serves L4Edge · Tier 1one per line, or per plantReads machines · buffersEdge · Tier 1the next lineReads machines · buffersSCADA / HMIreads the edge's UNSPLCsS7 · ModbusCNCOPC UARobotsEtherNet/IPSensorsMQTT · RESTnative protocolsMQTTMQTT · outboundNATS · Kafka · MQTTREST · SQL · IDocheartbeat

How to read the drawings​

The bands run from the controllers at the bottom (L1) to enterprise and cloud systems at the top (L5), with the plant DMZ as L3.5. A dark box is MaestroHub; its rows list what runs on that machine, from the application to the operating system. A dashed box is your own tooling, which MaestroHub works beside but does not ship. A red box is something customers often draw that MaestroHub does not provide today. Green lines read machines, orange lines carry data (the small orange block marks a buffer on disk), and purple dashed lines are management traffic to Fleet Manager. Plant names, machines and counts are examples.

How many servers​

LayoutHow manyWhatWhere
1. Line-side edge1 per plantTier 1L3
2. Edge plus DMZ hub1 per line, plus 3Tier 1 edges, one Tier 3 hubL3, L3.5
3. One hub, many sites1 or 3 in totalTier 2 (about 1 to 10 sites) or Tier 3; nothing at the sitesL4
4. Independent plants1 per plant, plus 1Tier 1 or Tier 2 per plant, Fleet ManagerL3 or L3.5, central
5. Three-layer namespace1 per line, 1 per plant, 3, plus 1Tier 1 edges, Tier 2 site hubs, a Tier 3 group hub, Fleet ManagerL3, L3.5, L4, central

Side by side​

1. Line-side edge2. Edge plus DMZ hub3. One hub, many sites4. Independent plants5. Three-layer namespace
Keeps running without the WANYesYesNoYesYes, at every layer
Survives a server failureNoThe hub doesWith Tier 3NoThe group hub does
Inbound connection into OTNoneNoneOnly for sources the hub pollsNoneNone
Kubernetes neededNoIn the DMZCentrallyFor Tier 2 plantsAt sites and centrally
Namespace across plantsNoNoYesNoYes

The layouts mix. A group can run layout 2 in its largest plant, add light sites to a central hub as in layout 3, and watch every deployment from one Fleet Manager as in layout 4.