Hi all,

I'm doing a spec for a financial site that will include setting up an
in memory database to hold in all equity trading data for the day and
would appreciate some insight.

The database is currently sql server though I'm seriously considering
moving to a linux / mysql as i've heard nothing but glowing reviews on
that.

The dataset would be about 6 gigs with about 33 million messages per
day sent to the imdb. If it wasn't for the license fees, I'd probably
look at splitting up the imdb to separate servers - ie one for each
stock exchange, to keep the dataset smaller and have room for
expansion.

I'm chatting with the Oracle team about TimesTen right now and that
seems to be the most robust, though the licensing fees are of course
painful.

What are your thoughts about TimesTen versus eXtremeDB, or one of the
open source once such as FastDB or MonetDB? How does mysql perform
with extremely large datasets? Are there any other standouts in this
field that deserves consideration?

Thanks for your input.

Joe