MaestroHub and Microsoft Fabric

Land plant data in a Fabric lakehouse through OneLake, or stream it through Event Hubs into an Eventstream, and read lakehouse files back to the line.

Connectors

Microsoft Fabric (OneLake) · Azure Event Hubs

Start with

A · Parquet files in OneLake

Network

Outbound HTTPS 443 only

The Microsoft Fabric (OneLake) connector writes, reads and lists files in a lakehouse, signed in with Microsoft Entra ID. It writes files, not tables: a notebook, a pipeline or the Load Table API turns them into Delta tables. For freshness in seconds, Azure Event Hubs feeds an Eventstream.

Find your route

Answer a few questions; the drawing below follows your answer.

Where does the data go?

Into Microsoft FabricBack to the plant

How fresh must it be?

SecondsMinutes are fine

What can the plant reach?

Fabric directlyOnly our Event Hubs

BThrough Event Hubs. Send to the Eventstream's custom endpoint, which gives an Event Hubs connection string, or to your own event hub.

Architecture

Context Engine runs next to your machines, and every route starts or ends there. Select a route to trace it.

Into Microsoft Fabric

Back to the plant

• Set up in MaestroHub
• Your infrastructure
• Set up in Microsoft Fabric

ALand files in OneLake. Minutes, as often as you load. Fabric capacity while the load runs. What each side sets up

Routes

A

Into Microsoft Fabric

Land files in OneLake

Start here for machine data

Freshness

Minutes, as often as you load

Microsoft Fabric compute

Fabric capacity while the load runs

Delivery

One file per batch; a load reads only the files it names

In MaestroHub

1. Batch rows with a Buffer or Aggregator node so each call carries many rows.
2. Add Convert to File with format Parquet and a declared column schema.
3. Write with OneLake Write File under Files/, Input Format base64.
4. To turn the files into a table from the same pipeline, call the Fabric Load Table API from a REST connection.

In Microsoft Fabric

1. A workspace and a lakehouse, with the service principal added to the workspace as Contributor.
2. A notebook, a pipeline or Load to Tables that converts Files/ into a Delta table. A bare .parquet file under Tables/ is not a table.

Choose it when Plant data should land in the lakehouse and minutes are fine.

Look elsewhere when You need seconds: use B.

OneLake guideConvert to File

B

Into Microsoft Fabric

Through Event Hubs

For second-level freshness

Freshness

Seconds

Microsoft Fabric compute

Fabric capacity runs the Eventstream

Delivery

At least once; remove duplicates on a message ID if needed

In MaestroHub

1. Create an Azure Event Hubs connection, with a SAS connection string or Microsoft Entra ID.
2. Use Send Batch per data set and keep store-and-forward on.

In your infrastructure

1. An event hub with its own consumer group for Fabric, unless you send to the custom endpoint.

In Microsoft Fabric

1. An Eventstream with the event hub, or a custom endpoint, as its source.
2. A destination: an Eventhouse KQL table or a lakehouse table.

Choose it when Freshness in seconds matters.

Look elsewhere when Minutes are fine: A uses no capacity between loads.

Azure Event Hubs guide

C

Back to the plant

Read files back

Close the loop on a schedule

Freshness

Your schedule

Microsoft Fabric compute

Only the notebook that writes the results

Delivery

Reads current files; safe to repeat

In MaestroHub

1. Pipeline: Schedule trigger, then List Paths and Read File, then publish to a UNS topic.

In Microsoft Fabric

1. A notebook that writes results as JSON or CSV under Files/results/.

Choose it when Fabric produces values the plant should use.

Look elsewhere when You want a SQL query against the warehouse: not available yet, see the note on this page.

OneLake guideSchedule trigger

No SQL read-back yet

The SQL Server connector signs in with a username and password, and Fabric's SQL analytics endpoint accepts only Microsoft Entra ID, so a pipeline cannot query a Fabric warehouse or lakehouse with SQL today. Read results back as files (C).

Compare the routes

Routes combine: most plants run one route in and one route back.

Route	Direction	Freshness	Microsoft Fabric compute	Delivery	Network	Best for
	Into Microsoft Fabric	Minutes, as often as you load	Fabric capacity while the load runs	One file per batch; a load reads only the files it names	Outbound HTTPS 443 to OneLake	Telemetry into lakehouse Delta tables and Direct Lake reports
	Into Microsoft Fabric	Seconds	Fabric capacity runs the Eventstream	At least once; remove duplicates on a message ID if needed	Outbound to Event Hubs	Real-time dashboards in Real-Time Intelligence
	Back to the plant	Your schedule	Only the notebook that writes the results	Reads current files; safe to repeat	Outbound HTTPS 443 to OneLake	Model outputs and targets written as files

Related

• All data platforms
• Microsoft Fabric (OneLake)
• Azure Event Hubs
• Convert to File