Prepare Data · Topic 1 of 7
Data Connections and Data Discovery
Create Fabric data connections with cloud, on-premises and virtual network gateways, and discover data with the OneLake catalog and the Real-Time hub.
- 5 min
- Connections are managed in Manage connections and gateways and reused across dataflows, pipelines and models.
- Cloud sources need no gateway; on-premises sources need an on-premises data gateway; virtual network sources need a virtual network data gateway.
- Connection roles: Owner, User, User with resharing.
What it is
Before you can bring data into Fabric you need a connection: a saved definition of a source, its address and the credentials used to reach it. Connections are reused by Dataflow Gen2, data pipelines, copy jobs and semantic models.
Before you build anything new, you should also check what already exists. The OneLake catalog finds Fabric items such as lakehouses, warehouses and semantic models. The Real-Time hub finds streaming data and events.
How it works
Creating a connection
You create and manage connections in Settings > Manage connections and gateways, or inline when you pick a source in a dataflow, pipeline or copy job. A connection stores the connector type, the address (such as server and database), the authentication method and a privacy level.
The path to the source decides which kind of connection you need:
| Where the source is | Connection type |
|---|---|
| A public cloud service, such as Azure SQL Database with public access | Cloud connection, no gateway |
| On premises or in a private network you manage | On-premises data gateway connection |
| An Azure service reachable only inside a virtual network | Virtual network data gateway connection |
Authentication options depend on the connector. Common ones are organizational account (OAuth 2.0), basic user name and password, service principal and account key.
Sharing a connection
A connection has its own permissions. The Owner manages it, a User can use it in their items, and User with resharing can also share it with others.
Sharing a connection lets a colleague build a dataflow against the same source without knowing the password. It does not give them any rights in the source system beyond what the stored credential allows.
Discovering items with the OneLake catalog
The OneLake catalog lists the Fabric items you have access to across all workspaces. Its Explore view filters by item type, workspace, domain, owner, endorsement and tags.
Selecting an item opens a details pane with its description, endorsement, sensitivity label, lineage and, for data items, a list of tables and columns. From there you can open the item, explore its data or build on it.
The catalog is also available from Power BI Desktop, so report authors can pick a certified semantic model or a lakehouse without leaving the tool.
Discovering streams with the Real-Time hub
The Real-Time hub is the catalog for data in motion. It lists data streams already flowing in the tenant, such as Eventstream outputs and KQL database tables, and it offers connectors for new sources.
| You find or connect | Examples |
|---|---|
| Existing data streams | Eventstreams and KQL tables you can access |
| Microsoft sources | Azure Event Hubs, Azure IoT Hub, change data capture from Azure databases |
| Fabric events | Workspace item events, OneLake file events, job events |
| Azure events | Azure Blob Storage events |
From a stream you can preview data, route it with an Eventstream into an Eventhouse or lakehouse, or set an alert with Activator. Detailed Eventstream processing belongs to data engineering; for DP-600, know that the hub is where you discover and connect streams.
A worked example
An analyst must combine on-premises SQL Server sales data with a product lakehouse another team already built.
- An administrator installs an on-premises data gateway on a server that can reach SQL Server.
- The analyst creates an on-premises connection to the SQL Server database through that gateway, using a service account.
- In the OneLake catalog the analyst filters on Certified and finds the product lakehouse, so there is no need to load product data again.
- A Dataflow Gen2 reads the sales tables through the new connection and lands them next to the product data.
- The analyst shares the connection with a colleague as a User, so the colleague can maintain the dataflow.
When a question mentions a firewall, a private endpoint or an on-premises server, the answer almost always involves a gateway. Choose the virtual network data gateway for Azure sources inside a virtual network.
How the exam asks about this
- "A Dataflow Gen2 must read an on-premises SQL Server database." Create a connection through an on-premises data gateway.
- "Users must find certified lakehouses in the Finance domain." Filter the OneLake catalog by domain and endorsement.
- "Find the Azure Event Hubs streams available and route one into Fabric." Use the Real-Time hub and create an Eventstream from the source.
What to remember for the exam
- Connections are managed in Manage connections and gateways and reused across dataflows, pipelines and models.
- Cloud sources need no gateway; on-premises sources need an on-premises data gateway; virtual network sources need a virtual network data gateway.
- Connection roles: Owner, User, User with resharing.
- The OneLake catalog discovers Fabric items and filters by type, domain, endorsement and tags.
- The Real-Time hub discovers streams, Microsoft sources, Fabric events and Azure events.
Sharing a connection shares the stored credential's access. Use a service account or service principal with only the permissions the dataflow needs.
Common mistakes
Installing an on-premises gateway for a public cloud source. A cloud connection is enough when Fabric can reach the service directly. Gateways add maintenance.
Searching workspaces one by one for data. The OneLake catalog searches every workspace you can access and shows endorsement, so it is the expected answer for discovery.
Using the OneLake catalog to find event streams. Streams and events are discovered in the Real-Time hub. The catalog is for Fabric items.
Check yourself
Which gateway do you use for an Azure SQL database that is reachable only through a private endpoint?
A virtual network data gateway, which runs inside the virtual network and needs no server of your own.
Where does a user find certified semantic models from all workspaces they can access?
In the OneLake catalog, filtered by item type and endorsement.
What can you do from a stream in the Real-Time hub?
Preview it, route it with an Eventstream into a store such as an Eventhouse, or set an alert with Activator.
Topic 1 of 7 in Prepare Data
Test yourself on Prepare Data
The full Microsoft DP-600 question bank has 648 questions, with timed practice exams and an explanation for every answer.
This study guide is independent and is not affiliated with, endorsed by, or sponsored by Microsoft. Vendor names, certification names and exam codes are trademarks of their respective owners and are used for identification only.