> We've traditionally called it mega-database manager. The right way to
> implement it is as a separate provider under Y-valve / dispatch module l
> looping back through the Y-valve / dispatch module to access target
> databases. This allows it to access local, remote, and external
> databases through gateways. The meta-database manager should be able to
> handled both statically declared meta-databases, defined in terms of
> existing database, and dynamically declared sets of source databases.
>
> It would be a dreadful mistake to try to implement the mega-database
> inside a database engine. The structures, dynamics, and optimization
> strategies are all wrong. A composite engine would be functionally
> limited, a bad database engine, and a bad mega-database manager.

Can you explain more your idea ? You mentioned it before but i didn't
understand it nor that time nor now.


Thanks,
Vlad