Interesting, I just found the thread here in Automated Trading.
It has it's own data storage method which is far better than any "database" for 2 reasons.
1. Tick data is very small number of bytes. When stored as a "row" or "object" it incurs overhead in the database that is larger than the tick and drastically increasing the space required for the data which is already quite large anyway.
2. TickZOOM reads it's data direction from the file into memory as binary. So it's extremely fast. It can load millions of rows (ticks) of data in mere seconds. Every database solution tried can take drastically longer. Object databases are the fastest at between 10 to 30 minutes. Relational DBs can take hours or days.
Sincerely,
Wayne
EDIT: The TickReader and TickWriter have source code so if someone wants to try other storage methods, be my quest. We're planning, at some point, to either use TREE or something like it for more complete solution to tick storage.