Forum Discussion
Columnar database
You should always use the same bitness of Power BI as you are using for Office. That said, you might be able to install the 64bit version of the Microsoft Access 2010 Redistributable engine. That will allow Power BI 64bit to read your ACCDB files. But be forwarned, I cannot guarantee that installing that 64bit package will not mess up your 32bit version of Office!
As to Power BI retaining history and only getting new data, that is exactly what the incremental refresh feature does, but it requires a backend database that supports query folding (Oracle, SAP, SQL Server, PostGRE SQL, etc, and possibly Access) but it will not work with any flat file of any sort as there is no engine on the backend to limit the data. Power Query must process all data to know what to keep and not keep, so it is still doing 95% of the work in just the query. Loading takes comparatively little time at all. So if you must stick with CSV files, just load it each time.
Note that there is an interesting workaround in this blog by Chris Webb, but I am not sure I would use it in production as you will constantly have to keep up with what is "loaded" and what isn't and get invested in the use of ALM Toolkit to manage the data in the cloud. To me it is more interesting as an exercise than useful as a production tool.
Good night. I first put this question in English and it was translated into Spanish. Interesting! I appreciate your answer. What happens is that the Power BI I use is 64 GB and the MS Access I use is 32 GB. And that forbids direct connection. I know that having all the CSVs in the same file allows you to import each CSV as long as they have their data in the same shape (columns with the same characteristics).
But if it's hundreds of text files each with millions of lines, I'm wasting time reimporting the data that's already been in Power BI, every time there's a new csv. That's why I thought I'd ask if it was possible to have one BI database history and another with new data and just add what's new to history. But what you're telling me isn't typical of using two distincto databases in the power BI model.
Again I appreciate your response and continue to investigate.
Community ChampionYou should always use the same bitness of Power BI as you are using for Office. That said, you might be able to install the 64bit version of the Microsoft Access 2010 Redistributable engine. That will allow Power BI 64bit to read your ACCDB files. But be forwarned, I cannot guarantee that installing that 64bit package will not mess up your 32bit version of Office!
As to Power BI retaining history and only getting new data, that is exactly what the incremental refresh feature does, but it requires a backend database that supports query folding (Oracle, SAP, SQL Server, PostGRE SQL, etc, and possibly Access) but it will not work with any flat file of any sort as there is no engine on the backend to limit the data. Power Query must process all data to know what to keep and not keep, so it is still doing 95% of the work in just the query. Loading takes comparatively little time at all. So if you must stick with CSV files, just load it each time.
Note that there is an interesting workaround in this blog by Chris Webb, but I am not sure I would use it in production as you will constantly have to keep up with what is "loaded" and what isn't and get invested in the use of ALM Toolkit to manage the data in the cloud. To me it is more interesting as an exercise than useful as a production tool.