The primary difference is in the systems original intention. The "old" database systems provided indexed (i.e., fast) access and very little overhead. They were designed for maximum speed. "Relational" database systems are primarily concerned with the integrity of the data first and speed second. They enforce referential integrity and provide commit and rollback features (i.e, transactions) and are generally logged for recoverability.

Anyone can create on offset to a file to quickly find information, but it takes a heck of a lot more than that to have a recoverable, high integrity data base management system.

I hope this helps.

Troy