Currently large datasets are cumbersome when trying to utilize drill by, as the list is very long and some columns are not useful to drill by.
This is currently solved by editing the dataset and un-checking "Is Dimension". This is not optimal as this also removes the option to utilize the column in chart creation. The other options is to create a virtual dataset with fewer columns however this may be less performant.
Suggested solution is to make the drill by column list both configurable at the dataset level with a new setting specifically for Drill By and configurable at the chart level by specifically setting what dimensions to filter by for that chart.