Aggregations (Public Preview)

Important

This content is archived and is not being updated. For the latest documentation, see Microsoft Dynamics 365 product documentation. For the latest release plans, see Dynamics 365 and Microsoft Power Platform release plans.

Note

These release notes describe functionality that may not have been released yet. To see when this functionality is planned to release, please review Summary of what’s new. Delivery timelines and projected functionality may change or may not ship (see Microsoft policy).

Massive volumes of data require new ways of storing information to balance the needs of slice-and-dice interactive analysis with deep, detail-level reporting. Aggregations allow model developers to surface cached values at a high level for interactive analysis, but still let users drill down to detailed data that is queried from the underlying data.

You can create DirectQuery models over massive-scale data sources such as Spark clusters, or massive data warehouses. For interactive analysis, running queries directly against these datasets is impractical. But for datasets that could be as large as hundreds of terabytes, the data cannot all be cached in memory. Aggregations lets you cache just aggregate data into memory for fast access. You define tables in your data model as an aggregate table, linked to tables at the detail level.

The detail tables stay in DirectQuery mode but the aggregates are defined as being in dual mode so the data is also cached in memory at the aggregate level. If users run queries or create visuals that can be answered from the in-memory cache, the results are retrieved from there. But if the query requires the detail data, it’s pushed down to the underlying DirectQuery source dynamically. The end user doesn’t see any difference in experience in their Power BI report.