Power Bi speed

[Umidjon]

Hi, 

I have faced with difficulities while using Power Bi Pro version. In our company we use Microsoft Access Database. And I connected Power Bi to Microsoft Access Database. Our data is so big(more than 1 million rows). It takes time when I import data from Access to Power Bi. In addition, power query performs very slowly. I know that Power Bi has some limitations. It cannot work with huge data. My question is how can I improve power bi speed? If I change my Access database to MySQL, the speed will increase or it does not change ? Do you know any ways which overcome my issue ? 

Please help !

Thank in advance 

[mahoneypat · Accepted solution]

Other sources could be faster for sure (and would give you the option to use Direct Query too), but you could also consider taking the refresh offline by publishing your model and setting up scheduled refresh on the dataset.  That way you can connect to the published dataset, especially now that we have composite models for datasets so you can add in other data sources if needed.

Regards,

Pat

[v-robertq-msft · Accepted solution]

Hi, @Umidjon 

According to your description, you said that your data source has more than 1 million rows. Import mode is not very suitable for you in this situation. My suggestion is to try to use the direct query mode to connect to Power BI or composite models that contain import and direct query. Of course, composite models can be the best way to achieve a balance between refresh speed and query time.

The fundamental approach to composite models is to set import mode for dimension-type tables and set DirectQuery fact-type tables from the same source. You can learn from this document to get the composite model guidance in Power BI Desktop:

https://docs.microsoft.com/en-us/power-bi/guidance/composite-model-guidance

One example of using composite models in Power BI Desktop

Best Regards,

Community Support Team _Robert Qin

If this post helps, then please consider Accept it as the solution to help the other members find it more quickly.

[mahoneypat · Accepted solution]

Other sources could be faster for sure (and would give you the option to use Direct Query too), but you could also consider taking the refresh offline by publishing your model and setting up scheduled refresh on the dataset.  That way you can connect to the published dataset, especially now that we have composite models for datasets so you can add in other data sources if needed.

Regards,

Pat