Most teams end up shipping their lakehouse data twice: once into Delta or Iceberg, then again into a relational database so Power BI and Tableau can render a chart. The DeltaForge ODBC driver removes the second copy entirely. It reads tables in your own storage, with no Databricks cluster behind it.
Signed binaries for every platform: deltaforge-org/delta-forge on GitHub
Signed builds for Windows, Linux, and macOS are attached to every delta-forge-odbc release. The driver registers under the canonical name DeltaForge ODBC Driver, so every ODBC client on the machine discovers it without further configuration.
Quickest is the one-line installer: it fetches the latest signed MSI, registers the driver, and drops the Power BI connector into place. Or download the MSI and run it yourself from an elevated shell. Then open the 64-bit ODBC Data Source Administrator (odbcad32.exe in System32) to add a DSN. Modern Power BI and Excel are 64-bit.
iwr -useb https://deltaforge.org/install.ps1 | iex; Install-DeltaForge -Pkg deltaforge-odbcmsiexec /i deltaforge-odbc-<version>-windows-x64.msi /qn /norestartQuickest is the one-line installer, which extracts the driver and registers it with unixODBC. Or do it by hand: extract the portable tarball, put the shared object on the library path, and register it with unixODBC. Verify with odbcinst -q -d, then add a DSN to ~/.odbc.ini.
curl -fsSL https://deltaforge.org/install.sh | sh -s -- --pkg deltaforge-odbctar -xzf deltaforge-odbc-<version>-linux-x64.tar.gz
sudo cp libdeltaforgeodbc.so* /usr/local/lib/ && sudo ldconfig
sudo odbcinst -i -d -f /dev/stdin <<'EOF'
[DeltaForge ODBC Driver]
Description=DeltaForge ODBC Driver
Driver=/usr/local/lib/libdeltaforgeodbc.so
UsageCount=1
EOFQuickest is the one-line installer. Or install the signed universal .pkg yourself. Either way it registers the driver with iODBC, the macOS-native driver manager, so Tableau and other iODBC clients pick it up.
curl -fsSL https://deltaforge.org/install.sh | sh -s -- --pkg deltaforge-odbcsudo installer -pkg deltaforge-odbc-<version>-macos-universal.pkg -target /Replace <version> with the current release (for example 1.0.610). Power BI users can also drop the bundled DeltaForgeOdbc-<version>.mez connector tile into Power BI's Custom Connectors folder. Setup, DSN locations, and the full connection-string reference live in the driver documentation.
BI tools want ODBC. Lakehouses historically offered notebooks and SDKs. The path of least resistance was to copy data into a relational database. DeltaForge closes that gap with a native ODBC driver.
Delta in cloud storage plus the same data in a relational store: two bills, two formats, two failure modes. Removing the copy removes one of each.
When Power BI reads Delta directly through the driver, there is no refresh window. The number you see is the number that exists in the table right now.
RBAC and the audit log are enforced at the ODBC layer by the same engine that governs every other access path. No second GRANT list in a separate database to keep aligned.
A driver that looks like every other ODBC driver to the tool on the other end
Standard ODBC, so the tool only needs to know how to speak ODBC. No bespoke connector, no plugin, no managed gateway.
Passwords and tokens are kept in the operating system keychain, not in a plaintext config file. Corporate proxies are supported.
Decimals, timestamps, unicode strings, and exact column names land in the BI tool the same way they exist in the Delta table.
Iceberg tables are first-class in the DeltaForge engine, so the ODBC driver queries them the same way it queries Delta
An Iceberg table and a Delta table sit side by side in the same catalog and answer the same SQL. The BI tool never needs to know which format is underneath.
Iceberg metadata is read natively by the engine. There is no Spark cluster or JVM gateway between the spreadsheet and the table. See Apache Iceberg support for the full read and write surface.
The same engine that serves the driver runs INSERT, UPDATE, DELETE, and MERGE on Iceberg, so the tables your dashboards read are maintained over the same connection.
Short answers to the questions teams ask before pointing a BI tool at the lake
Yes. This driver reads Delta tables in your own storage (S3, ADLS, or local disk) through the DeltaForge engine. There is no Databricks cluster, SQL warehouse, or other managed compute behind the connection.
Yes. The same driver queries Iceberg tables alongside Delta tables through the same DSN, so Power BI, Excel, Tableau, and DBeaver query them like any other database table.
Power BI Desktop and Service, Tableau Desktop and Server, Excel, DBeaver, SQL Workbench, .NET via System.Data.Odbc, Python via pyodbc and turbodbc, R via RODBC and odbc, and unixODBC isql.
How the driver performs and what the engine behind it can do to the tables it serves
Measured throughput of the ODBC and ADBC drivers under the access patterns Power BI, Tableau, and .NET actually generate.
The same engine that answers your dashboards also maintains the Delta tables they read, in plain SQL.
Full DML on the Iceberg tables this driver exposes, with no JVM anywhere in the stack.
Install the driver, point Power BI at Delta, and delete the pipeline that was only there to feed a chart.