- disk space conservation (normalization provides for this)
- common data access language (SQL)
- easily exchangeable resources - DBA/DBD's (not cheap, but interchangeable)
- common/standard backup/recovery (data security)
And the downside:
- scalability (I think they could scale hugely, but accessing would prohibit google size type of needs)
As with most great ideas, the original intention/benefits are often reduced through use of the idea in areas not intended or thought of. I think a prime area where databases fail are large size websites where quick (very quick) access to massive data is required (not reporting or transaction processing). Years ago large (considered large at that time) data stores existed outside of databases, standard summarizing was used to gather specific info, feed it into the database which would then be used for reporting....are we heading that way again?