[Sqlrelay-discussion] Is it possible to do mdb->sqlrelay->mdb scheme?

[if <ze...@gm...>]

Hello list!

I face a little challenge, that after some heavy googling seem to be
not so little. Here's the problem:

We have custom (*cough* goverment) application for windows, that works
with mdb files _only_. It runs on windows, it opens only predefined
mdb file, and it is dumb and enerprisey, and the worst part -- we are
locked to use it. There's no security in it, no logging, and no
control over the user. It works with financial data, so i have to
tighten it a bit.

I moved mdb file to linux samba server, and restricted access on
network level. The next part is tricky. I need control at sql query
level and make an alert system, because main threat is emloyees
working with it. My plan is to inject sqlrelay between "fake" mdb file
and some good database (think postgres). Or at least inject it between
fake mdb and real mdb, and log sql, that will be sufficient for the
first time. After that i'll write everything i need.

So the main question is -- how can i make this fake mdb file? Sqlrelay
can make socket->mdb proxying, but can it make a reverse proxy? Than
i'll just chain few instances together and get what i want.

I also looked at Perl::DBI Proxy guides, mdbtools guides -- all with no luck.
Any, just any advice will be highly appreciated!


Thanks for reading,
regards,
if