I've tried adding a non-clustered index to some tables and while it has improved performance significantly, the index takes up 5-10 times more space than the data in the table. It's a rather simple index:
CREATE NONCLUSTERED INDEX [Product_ArrivalID]
ON [dbo].[Product] ([ArrivalID])
It only has one column and four include columns and yet it's taking up so much space. What is the best practice in this type of scenario? The problem is that I need one such index on every table in order to speed up an ETL load that queries staging tables.