Query caching in Power BI Premium/Embedded

Organizations with Power BI Premium or Power BI Embedded can take advantage of query caching to speed up reports associated with a dataset. Query caching instructs the Premium/Embedded capacity to use its local caching service to maintain query results, avoiding having the underlying data source compute those results.

Important

Query caching is only available on Power BI Premium or Power BI Embedded. It is not applicable to LiveConnect datasets leveraging Azure Analysis Services or SQL Server Analysis Services.

Cached query results are specific to user and dataset context and always respect security rules. At present the service only does query caching for the initial page that you land on. In other words, queries aren't cached when you interact with the report. The query cache respects personal bookmarks and persistent filters, so queries generated by a personalized report will be cached. Dashboard tiles that are powered by the same queries also benefit once the query is cached. Performance especially benefits when a dataset is accessed frequently and doesn't need to be refreshed often. Query caching can also reduce load on your Premium/Embedded capacity by reducing the overall number of queries.

You control query caching behavior on the Settings page for the dataset in the Power BI service. It has three possible settings:

  • Capacity default: Query caching Off

  • Off: Don't use query caching for this dataset.

  • On: Use query caching for this dataset.

    Query caching dialog box

Considerations and limitations

  • When you change caching settings from On to Off, all previously saved query results for the dataset are removed from the capacity cache. You can turn off caching either explicitly or by reverting to capacity default setting that an administrator has set to Off. Turning it off can introduce a small delay the next time any report runs queries against this dataset. The delay is caused by those report queries running on demand and not leveraging saved results. Also, the required dataset may need to be loaded into memory before it can service queries.
  • When the query cache is refreshed, Power BI must run queries against the underlying data models to get the latest results. If a large number of datasets have query caching enabled and the Premium/Embedded capacity is under heavy load, some performance degradation may occur during cache refresh. Degradation results from the increased volume of queries being executed.

Next steps