I have a table with four columns and ~660 million rows. Two of the columns are bigint although they are not really required to hold any content that could not also fit into int. So as a test I changed their datatype from bigint to int. There is a clustered index on that table which uses neither of those two columns for its key.
The result is puzzeling to me. The data space (in the table properties) stayed exactly the same while the index space dropped 23.5%. Can someone explain to me what happened there? Why did the data space not change?
As an aside, I did something similar to another table. There the data space dropped 30,4% which is exactly what I calculated from the change in row size. The drop in index space was similar.
[edit note: I originally overlooked restoring one nonclustered index. This has been fixed and the question changed accordingly]