There is obviously a four byte per row space savings in a data column when using INT vs a BIGINT type.
But I have been trying to figure out if there is also any space savings for myisam indexes on integers since they are typically unpacked.
MySQL doesn't say what the structure type is for indexes - it only calls the units "nodes" and then doesn't mention their actual size.
So given the same set of numbers, say a million rows of integers, does using INT instead of BIGINT reduce the node size? Or because there can still be an unlimited number of rows despite the integer range, does the node size always have to be rather large?
I guess I could whip up a demo to test this question but maybe someone already understands this properly on a technical level and can explain it.