Compare
XLTable compared with SSAS, Power BI, Cube, AtScale, icCube and Mondrian
XLTable does one job: native Excel PivotTables, a pivot table in the browser and AI assistants on the warehouse you already have — ClickHouse, Snowflake, BigQuery, Databricks, StarRocks, Trino, Greenplum or DuckDB — with no copy of the data. Here is how that compares with the tools teams usually look at.
Facts about other products were checked against their own documentation in November 2026; the links are under each section.
At a glance
Swipe the table sideways to see the other products →
| XLTable | SSAS | Power BI / Fabric | Cube | AtScale | icCube | Mondrian | |
|---|---|---|---|---|---|---|---|
| Native Excel PivotTable (built-in Analysis Services connection) | ✓ No add-in, hierarchies and drill-down | ✓ | ✓ Analyze in Excel; hierarchies only on Import-mode models | Preview — Enterprise plan, views only | ✓ | ✓ Some PivotTable features not yet implemented | Not documented by Pentaho; ✓ in the eMondrian fork |
| Where the numbers are computed | In your warehouse — every PivotTable action is one SQL query, no copy | Usually in the cube — a copy of the data, processed on a schedule | Import copies the data; DirectQuery and Direct Lake query it | In the warehouse; pre-aggregations recommended for MDX | In the warehouse, plus aggregate tables AtScale builds there | In icCube memory — data is loaded into the server | SQL in the database; MDX calculations and caches in the Java server |
| ClickHouse, StarRocks, Greenplum, Trino, DuckDB | ✓ All five, live | Imported into the cube | ClickHouse and Trino through partner connectors | ClickHouse, Trino, DuckDB | None of the five | Through JDBC, loaded into memory | Greenplum; ClickHouse in eMondrian |
| Where it runs | Linux or Windows, on-premises or your cloud, air-gapped | Windows with SQL Server | Microsoft cloud tenant; Report Server has no Analyze in Excel | Cube Cloud; the MDX API is not in open-source Cube Core | Self-hosted: Kubernetes or Linux, 16+ cores and 64 GB RAM | Self-hosted Java server or Docker | Self-hosted Java server |
| Price | $30 per user / month from 50 users; free Desktop edition | Per SQL Server core | Per user, plus Fabric capacity for the XMLA endpoint | Enterprise plan — on request | On request | Embedded license from $1,950 / month | Pentaho: commercial license for production |
What XLTable is
- • An XMLA-compatible OLAP server: Excel connects through the standard Analysis Services dialog, no add-in.
- • Every PivotTable action becomes one SQL query in your warehouse, with a shared result cache. XLTable keeps no copy of the data.
- • The cube is one SQL file kept in Git. The same cube serves Excel, a pivot table in the browser, AI assistants over MCP and, in the server edition, BI tools with a PostgreSQL driver over the SQL endpoint — with one set of roles and row-level security.
- • Runs on Linux or Windows, on-premises or in your cloud, air-gapped if needed.
XLTable vs SSAS
Microsoft Analysis Services is where most Excel teams come from. XLTable gives the same PivotTables, hierarchies and MDX on the warehouse you have now — without a cube to process.
- • In the usual setup SSAS keeps its own copy of the data in the cube and refreshes it with processing jobs. XLTable keeps no copy: every PivotTable action is one SQL query in the warehouse, so the numbers are as fresh as the warehouse.
- • Microsoft’s list of data sources for multidimensional models names SQL Server, Oracle, Teradata, Db2 and a few others, and ODBC sources are not supported; ClickHouse, Snowflake and BigQuery are not on it. XLTable queries them live.
- • An SSAS cube is a Visual Studio project. An XLTable cube is one SQL file — measures, dimensions, hierarchies and access rules — kept in Git.
- • SSAS runs on Windows with SQL Server and is licensed per SQL Server core. XLTable runs on Linux or Windows, on-premises or in your cloud.
- • For users nothing changes: Excel connects through the same Analysis Services dialog, with roles, row-level security and Active Directory. The same cubes also open in a pivot table in the browser and in AI assistants.
Sources: SSAS supported data sources, Processing a multidimensional model
SSAS replacement: the full comparison and how a migration goes →XLTable vs Power BI and Microsoft Fabric
Not a replacement — a complement. Keep Power BI for dashboards; use XLTable where analysts need live PivotTables on a warehouse outside the Microsoft stack.
- • Hierarchies: Microsoft documents that only Import-mode semantic models keep their hierarchies in Analyze in Excel — models on DirectQuery, Direct Lake or composite models lose them. XLTable keeps hierarchies on live queries.
- • Size: Import mode copies the data into the model, with per-license size limits; Analyze in Excel may fail on data over 2 GB, and Microsoft points to the XMLA endpoint, which needs Premium, Premium Per User or Fabric capacity.
- • Tenant: users outside the Microsoft Entra tenant that hosts the model cannot connect from Excel, and Power BI Report Server — the on-premises option — does not support Analyze in Excel. XLTable is installed in your network and works with Active Directory or local accounts.
- • Sources: Direct Lake reads Delta tables in OneLake. XLTable queries ClickHouse, StarRocks, Greenplum, Trino and DuckDB live, where they are.
Sources: Analyze in Excel, Direct Lake overview, XMLA endpoint
XLTable vs Cube
A semantic layer built for applications, with Excel as an add-on. In XLTable the Excel PivotTable is the core of the product.
- • Cube’s MDX API — the native Excel PivotTable connection — has been in preview since July 2024 and is enabled by Cube’s account team.
- • It requires the Enterprise plan and the Performance M deployment tier of Cube Cloud, works only with views, and is not available in open-source Cube Core. XLTable’s free Desktop edition includes the same Excel connection as the server.
- • For sub-second MDX, Cube recommends routing queries to pre-aggregations. XLTable pushes every PivotTable query to the warehouse as SQL and caches the results.
- • Cube lists ClickHouse, Trino and DuckDB among its sources, but not StarRocks or Greenplum.
Sources: Cube MDX API, Cube and Excel, Cube pricing
XLTable vs AtScale
A mature enterprise semantic layer for the big cloud warehouses. XLTable is lighter, SQL-first and covers the open-source engines AtScale does not.
- • AtScale supports Snowflake, Databricks, BigQuery, Redshift and PostgreSQL — not ClickHouse, StarRocks, Greenplum, Trino or DuckDB.
- • Self-hosted AtScale needs Kubernetes or a Linux server with at least 16 cores and 64 GB of RAM. XLTable runs on an ordinary Linux or Windows server, and most installations take under an hour.
- • AtScale prices are not published. XLTable is $30 per user per month from 50 users, and the Desktop edition is free.
- • The Snowflake XMLA endpoint powered by AtScale, announced in June 2026, is listed as “private preview coming soon”; time intelligence, hierarchies and dimensional modeling are AtScale Enterprise features.
Sources: AtScale supported platforms, AtScale system requirements, AtScale and Snowflake Semantic Views
XLTable vs Databricks Excel Add-in and metric views
Bundled with Databricks. The difference is depth: XLTable gives a live native PivotTable, the add-in imports results into a sheet.
- • The add-in builds the pivot in its own pane and imports the result into a new sheet; a refresh clears and reloads the data and removes columns you added. XLTable is a native PivotTable with drill-down, slicers and hierarchies on every click.
- • Unity Catalog metric views keep improving — window measures and fiscal-period offsets became generally available in October 2026. XLTable reads the same Databricks tables and also ClickHouse, StarRocks, Greenplum and the other supported engines with one cube format.
- • The same XLTable cube serves Excel, a pivot table in the browser, AI assistants over MCP and, in the server edition, BI tools with a PostgreSQL driver over the SQL endpoint, with one set of roles.
Sources: Databricks Excel Add-in, Metric views: advanced techniques
XLTable vs icCube
An in-memory OLAP server focused on embedded dashboards. XLTable keeps no copy: the data stays in your warehouse.
- • icCube builds an in-memory cube: data is loaded from the sources into the icCube server and refreshed incrementally. XLTable sends every query to the warehouse, so the numbers are as fresh as the warehouse and the cube size is not limited by server memory.
- • Both connect to Excel natively over XMLA; icCube’s documentation notes that some PivotTable features are not yet implemented.
- • icCube’s embedded license starts at $1,950 per month; there is no free edition, only a trial. XLTable Desktop is free.
Sources: icCube server, icCube and Excel, icCube pricing
XLTable vs Mondrian and eMondrian
The classic Java OLAP engine behind Pentaho, and its fork with Excel support. XLTable defines the cube in SQL and leaves the computing to the warehouse.
- • Mondrian sends aggregation queries to the database, but MDX calculations, dimension member lists and its segment cache live in the Java server’s memory. Pentaho’s own tuning guide recommends aggregate tables for large warehouses and warns against in-memory cross joins. XLTable turns each PivotTable action into one SQL query.
- • Cubes are XML schema files edited in Schema Workbench. An XLTable cube is one SQL file — measures, dimensions, hierarchies and access rules — kept in Git.
- • Mondrian is published under the Business Source License: production use needs a commercial license from Pentaho.
- • Pentaho does not document Excel’s built-in Analysis Services connection for Mondrian. The eMondrian fork does and adds a ClickHouse dialect; per-user access for its AI agents comes only in the Enterprise edition. In XLTable, AI assistants see exactly what the user may see, in every license.
Sources: Mondrian native evaluation, Pentaho Analyzer performance tips, Mondrian on GitHub (license), eMondrian and Excel, eMondrian editions
See it on your own data
Connect Excel to the public demo server, or install the free desktop edition and point it at your warehouse.