Subject High selectivity
Author Jonathan Neve

Everyone always says that one should avoid indices with high
selectivity, and that they can even make the query slower.
I understand that.
But doesn't that depend on how evenly the values are distributed?
Suppose for instance that I have a STATE field on a table, that has two
possible values, 0 and 1.
All new records start out in STATE 0, and after a while, they end up in
STATE 1. Obviously, the number of records in STATE 0 will remain roughly
constant, whereas the number of records in STATE 1 will be continually
growing. If (as it is highly likely), I know in advance that I will
almost never need to select all records in STATE 1, but very frequently
all records in STATE 0, would it not be advantageous to define an index?


Jonathan Neve.