I recently found a table where I work and they have a cluster index that contains a date/time column. This column contains date (year, month, day), and time (hour, minute, second, millisecond).
I can't see how that would ever be useful or not hinder the performance of the table instead of helping it. Most lookups to that table involve its primary key (which is not a date/time, it is an identity integer), second most used lookup involves another integer column which is a foreign key reference.
Thoughts? Can any good come from having a date/time in a clustered index?