I suspect that depends on the particular linguistic / technical community.
Certainly distributed system programmers will speak / think of a database and a database server as completely different things; each server may host multiple databases, and any given database may be served by multiple servers at different (or the same) time(s).
A database being down and a server being down (for instance) are completely different ideas. The people that are authorized to access a database are likely a different set of people than those authorized to access the server(s) as such, and so on.
Don't know if that helps, or just introduced whole new levels of uncertainty. :)