perform worse than clustered indexes on ranged searches where the result set is large, since each result could cost an additional I/O-operation to get the actual data record
search for b># using non-clustered index on b
en.wikipedia.org — Index (database) - Wikipedia, the free encyclopedia Add comments