Power BI Data Refresh and Gateway Setup

Power BI data refresh re-imports your source data on a schedule so that published reports show current figures rather than a snapshot from build time. For cloud sources this often works out of the box, but the moment your data lives behind a corporate firewall, on an on-premises SQL Server, a network file share or a private database, you need the on-premises data gateway. The gateway is a small service that sits inside your network and gives the Power BI Service a secure, outbound channel to those sources without opening inbound ports. Getting refresh and the gateway right is what turns a one-off dashboard into a dependable daily decision tool.

Import versus DirectQuery: what actually refreshes

Before configuring anything, know your storage mode. With Import, data is loaded into the model and refresh re-pulls it on a schedule, which is what most reports use and what this guide focuses on. With DirectQuery, queries hit the source live, so there is no scheduled data refresh of the cached data, though the gateway is still needed to reach an on-prem source. Composite models mix both. The rest of this guide assumes Import unless stated.

Choosing and installing the gateway

There are two gateway types. The standard on-premises data gateway runs as a Windows service, can be shared across users and datasets, supports clustering for high availability and is the only sensible choice for production. The personal mode gateway runs under a single user account, cannot be shared and suits an individual analyst's own reports.

Install the standard gateway on a server that stays on, has a stable network path to your sources and is not someone's laptop. Sign in with an account that will own the gateway, register it, and note the recovery key safely. Once installed, you add data sources to the gateway in the Power BI Service under Manage connections and gateways, supplying the credentials Power BI should use to connect.

Mapping the dataset to the gateway

After publishing your report, open the dataset settings in the Service. Under Gateway connection, bind each data source in your model to a matching data source defined on the gateway. The source definitions must align: the same server name and database as used in Power BI Desktop. A frequent cause of refresh failure is a mismatch here, where the model points at a server name the gateway does not recognise.

Scheduling refresh and understanding the limits

With the gateway connected, set up Scheduled refresh in the dataset settings. You pick a frequency, time zone and the times of day to run, and optionally a failure notification email. The number of refreshes per day depends on your licence.

These figures are correct as of 2026, but Microsoft adjusts limits over time, so confirm current Microsoft licensing and refresh limits before you design around them. If you need near-real-time data, scheduled refresh is the wrong tool; consider DirectQuery, hybrid tables or streaming instead. There are also timeout limits, two hours on shared capacity and longer on dedicated capacity, that constrain how heavy a refresh can be.

Troubleshooting refresh failures

Most refresh problems fall into a few recognisable categories.

• Credential expiry: the stored credentials for a source have lapsed or a password rotated. Re-enter them in the gateway or dataset settings. This is the single most common failure.
• Gateway offline: the server hosting the gateway was rebooted, lost network, or the service stopped. Check the gateway status in Manage connections and gateways.
• Source unreachable or schema changed: a renamed column, dropped table or moved file breaks the query. Reconcile the model with the source.
• Timeout on large refreshes: the dataset exceeds the capacity time limit. Reduce data volume, use incremental refresh, or move to higher capacity.
• Data source mismatch: the model's connection string does not match the gateway data source definition. Align them exactly.

For large, slow datasets, incremental refresh is the most effective fix. Instead of reloading all history every time, it refreshes only a recent window of data and keeps older partitions untouched, cutting refresh time and load dramatically.

Good practice for reliable refresh

Run the gateway on a dedicated, always-on server and keep it updated; Microsoft releases monthly gateway updates. Use a service account rather than a personal login so refresh does not break when someone leaves. Turn on refresh failure notifications to a shared mailbox so problems surface quickly. Where possible, set up a gateway cluster so a single server outage does not stop every refresh. Finally, document which datasets depend on which sources so the impact of a planned source change is obvious.

Refresh reliability sits alongside a clean model and disciplined release process. If your reports also move between environments, our guide to Power BI deployment pipelines covers how to manage that safely, and our Power BI consulting service can take ownership of the whole pipeline.

If refresh failures, slow gateways or unclear schedules are eroding trust in your reports, a Trusted Numbers Review reviews your gateway, refresh and licensing setup in a fixed engagement and gives you a prioritised list of fixes. Book a review to make daily data something your team can rely on.