Incremental Refresh

[Mirkow1]

I just started using Incremental Refresh.
I have a data base with ## mill rows aprox, and wanted to use this feature. I created a file with only the mentioned base to try and see if this worked. Im using premium in my Power BI, logged obviously.

My base comes from SQL and i made a view that updates daily. I made the RangeStart, RangeEnd, change my date column as date/time, and made sure that it stays as query folding (I can access to the native query option in power query), then add the parameters. Then added the incremental refresh options for 7 days.
With another file using less data. the loading time without incremental refresh took 26 seconds, and with incremental it took 10 seconds. This was with half a dozen tryouts. I assumed that this was working and would work with my ## mill file.
The refresh time with the ##mill data takes around 30 minutes in refresh, and i would assume that with incremental refresh it would take less than that, taking in consideration that i dont have anything else on that file. But the loading time stays about the same with the tryouts i made.

Is there something im not taking into consideration?
Thanks by advance.

[lbendlin]

I just started using Incremental Refresh.
I have a data base with 29 mill rows aprox,

That's not really large enough to warrant incremental refresh, unless your database is really slow.

Note that Incremental refresh will reset every time you make a meta data change (like you probably did during your tryouts). 

Only use incremental refresh with immutable data coming from slow data sources, or when a partition size reaches 200 Million rows.

[Mirkow1]

What do you mean with "Meta data change"? And slow data sources?

[lbendlin]

Meta Data change = adding or removing columns

Slow data source = takes more than a minute to deliver 20 M rows.

[Mirkow1]

I haven´t added new columns and won´t be changing anything to the data base.
The issue is that it has at least 60 columns. My idea was that incremental refresh would diminish the amount of data that would be updating at the time to a segment of time (7 days in this example), taking into account that it wouldn´t update older rows from the date/time column with the parameters.