Initial dax query run against a Fabric semantic model is taking time

[Sreeja_G]

Hi,

Initial dax query run against a Fabric semantic model is taking time. what could be the reason for it and any optimization that can be done while retrieving data from Fabric to avoid taking more time in the intial run.

[v-huijiey-msft]

Hi @Sreeja_G ,

The reason for the initial DAX query may take the following points:

1. When the first access semantic model or data is accessed, the initial loading time is slow.

2. Data volume is too large.

3. Dax query statements are more complicated.

Regarding the optimization of DAX, you can consider from the following aspects:

1. Use the latest version of Power BI to take advantage of the latest performance enhancements.

2. Minimize the number of calculations and aggregations performed in your Dax queries.

3. Avoid using nested calculations in your Dax queries, as they can be resource-intensive.

4. If possible, use simpler Dax functions instead of more complex ones.

Where appropriate, use calculated tables to precalculate frequently-used data.

5. Minimize the use of calculated columns, as they can slow down query processing.

For more relevant information, please refer to:

How to Optimize Dax Queries in Power BI - Zebra BI

If you have other questions, please contact me at any time.

Best Regards,
Yang
Community Support Team

If there is any post helps, then please consider Accept it as the solution  to help the other members find it more quickly.
If I misunderstand your needs or you still have problems on it, please feel free to let us know. Thanks a lot!

[govindarajan_d]

Hi @Sreeja_G,

Does the DAX query use DirectLake mode or DirectQuery mode?

[Sreeja_G]

Yes @govindarajan_d it uses Direct Lake

[govindarajan_d · Accepted solution]

Hi @Sreeja_G,

There are many factors that can influence and we need to dive deeper to understand. Did you try using the Performance analyzer in Power BI to find out where DAX query spends lot of time?

1. Check if it falls back to DirectQuery mode. Check this blog: https://fabric.guru/controlling-direct-lake-fallback-behavior

2. Check if you have the correct capacity for running the queries and there are not so many parallel queries running corresponding to the capacity you use. 

In addition to this, you can try prewarming your dataset by calling Power BI REST API for running queries which will fill the cache. Check this: https://www.tackytech.blog/how-to-pre-warm-your-power-bi-semantic-model-from-fabric-data-pipelines/#...