A report in the service shows the data its semantic model had at the last refresh. A model that reads a cloud source can refresh on its own; one that reads a server inside the company network, such as an on-premises SQL Server, needs an on-premises data gateway to reach it.
In the semantic model's Settings:
- Gateway and cloud connections maps each data source to a connection on a gateway. Pick the connection that points at the same server and database the model reads.
- Scheduled refresh turns on with Keep your data up to date. Choose Daily or Weekly (and the days), the Time zone, and one or more times. A model in shared capacity can refresh up to 8 times a day.
Times are in the time zone you pick, so a 06:00 refresh for a team in New York needs the Eastern time zone, not UTC. Refresh only as often as the source changes: a model that refreshes every hour against data loaded once a night just adds load. Each section of the settings page has its own Apply button.
The source is mapped to the gateway connection for the same database, then refreshed twice a day.