Faster, Simpler Dashboards with Spice and Power BI

Spice AI built a Microsoft Power BI Connector on top of the Flight SQL ADBC driver that makes it easy for Power BI users to query across operational databases, analytical warehouses, and object stores. Spice federates large OLTP and OLAP datasets and accelerates them at the application layer in DuckDB and Arrow to support sub-second dashboards, driving significant performance and ease-of-use improvements for enterprise BI use cases. Under the hood, Spice leverages Arrow Database Connectivity (ADBC) and Flight SQL for Arrow-native performance that eliminates row-to-column conversion overhead.

Enterprises rely on Power BI to analyze and visualize data, but the data itself often lives across many systems: operational databases like Postgres or MongoDB, historical datasets in S3 or Delta/Iceberg tables, and real-time streams from systems like Kafka. Traditionally, making this data available in Power BI requires complex ETL pipelines, duplicated storage in various warehouses, and ongoing engineering effort to keep everything in sync. This results in dashboards with slow refresh times and fragile, operationally-intensive pipelines.  

Spice changes this dynamic by combining query federation and local acceleration into a single, lightweight runtime. Instead of copying data into a warehouse, Spice connects directly to distributed systems, federates queries across them, and accelerates datasets with DuckDB and Apache Arrow at the application layer. 

The Spice.ai Power BI Connector sits between Power BI and the systems it needs to query. Data federation removes the need to consolidate sources up front, and Power BI can treat them as if they were a single dataset; analysts can configure Spice once and immediately query across their data estate. For Power BI users, this means dashboards can run on live data from OLTP and OLAP sources without the burden of additional infrastructure management.

‍

‍

Federation simplifies connectivity, but acceleration delivers performance; the Spice runtime prefetches working sets of data from these upstream systems and locally accelerates them using DuckDB and Apache Arrow. This eliminates repeated network round-trips, stores data closer to applications, and enables sub-second queries - even for billion-row datasets. Data can be refreshed on a schedule or in real time with CDC, ensuring dashboards always present the latest information.

Further, unlike traditional lakehouses that focus on analytical workloads, Spice accelerates both operational and analytical data. Teams can query transactional sources like Postgres alongside large analytical datasets in S3, all with the same sub-second performance and without overwhelming production databases. For example, Spice can ingest row-based Postgres tables, accelerate them in DuckDB, and serve the results to Power BI in columnar Arrow format for interactive dashboards.

‍

Under the hood, the Spice.ai Power BI Connector is built on the Arrow Database Connectivity (ADBC) and Flight SQL. ADBC provides a vendor-agnostic, Arrow-native standard for delivering columnar data directly to applications and dashboards. In contrast to JDBC and ODBC, ADBC avoids row/column conversions and provides a faster, more efficient path for analytical queries.

Let's take this out of the abstract and introduce a practical use case to illustrate the value of Spice.

Consider an order management platform that previously relied on an AWS RDS database for recent transactions and S3 for longer-term history. Without Spice, a dashboard showing historical orders would require querying RDS directly, moving potentially millions of rows across the network, and joining that data with S3 data in a warehouse. With Spice, those tables can be federated, accelerated locally, and queried directly by Power BI. 

The video illustrates the performance delta between Spice's acceleration model (the dashboard on the right side of the screen) and a more traditional approach of querying RDS directly (the dashboard on the left side of the screen): query time is decreased from 1324 ms to 223 ms with Spice compared to RDS. That's the difference between a clunky analyst experience and an interactive, real-time workflow.

‍

‍

‍

API Key: Your Spice.ai API key for authentication (required for Spice Cloud). Follow the guide to obtain it from the Spice Cloud portal.

After establishing a connection, Spice datasets appear under their respective schemas, with the default schema being spice.public. When writing native queries, use the PostgreSQL dialect, as Spice is built on this standard.

For a list of supported data types, visit the docs here.

Spice extends Power BI beyond the traditional limits imposed on it by sub-optimal ETL pipelines.

Spice enables Power BI users to:

‍

To try it out, download the Spice Power BI Connector and follow the documentation. Configure your first dataset, connect Power BI, and experience how federation and acceleration can make your dashboards faster and more reliable.