I honestly don't like either of the three solutions. They don't really scale when you have a lot of combinations of attributes. I have deployed this type of da

I honestly don't like either of the three solutions. They don't really scale when you have a lot of combinations of attributes. I have deployed this type of data using both solution #1 and the Entity Attribute Value model and I honestly prefer the latter. We now have incorporated new data from multiple sources and the number of attributes has grown to around 1000. I don't see how deploying a table with 1000 columns can be considered a good solution. Also, the data is used downstream in multiple places which means that If I had gone with solution #3 (which looks like the best one) we would have needed to edit tens of queries to add a new join with the new table for the class with custom attributes