Fast insertion – what Tokutek excels at – isn’t only important for handling high data rates. It also enables a new way to extract value from an existing database called ad hoc indexing. KAYAK has a billion rows stored in TokuDB and adding an index for a traditional MySQL storage engine would take too long to be of any practical value. With TokuDB, however, an index can be created on the fly to accelerate any ad hoc query that would otherwise be causing a vastly slower full table scan.
Here are the details.