DAX Optimizer is designed to optimize Import models. However, it can also analyze DirectQuery models, but with some limitations:
-
Some of the issues detected may not be relevant for DirectQuery models. Check the issue documentation and description to figure out whether the proposed solution is specifically designed for VertiPaq and not for DirectQuery. When you do not see specific advice for DirectQuery, you can assume that the proposed solution is valid for both Import and DirectQuery models. However, the performance improvement may be lower for DirectQuery models. You should always test the proposed solution in your environment to evaluate the performance improvement.
-
The VPAX file might be not accurate for DirectQuery models. When you generate a VPAX file on a model that has DirectQuery tables of partitions (such as hybrid tables), it might not have all the required statistics. The lack of these statistics does not allow DAX Optimizer to correctly detect and evaluate issues in the model. Retrieving model statistics from a DirectQuery model can be a very time-consuming operation: for this reason, it could be disabled on tools that generate the VPAX file. For example, DAX Studio has a setting in Options / Standard / VertiPaq Analyzer called Read statistics from DirectQuery tables that must be enabled to collect statistics from DirectQuery tables.
NOTE: If your model contains Direct Query tables, ensure that the VPAX extraction tool is configured to include statistics for Direct Query tables. Missing or incomplete statistics may result in the DAX Optimizer being unable to analyze the model, and the uploaded VPAX file being reported as unsupported.