Hi,
I think that except if you already know you will need to scale or
use a
particular approach, you shouldn't worry about that while developping
your application, and just dump your data to be able to load it
back. I
think that's one of the great things with Smalltalk, you can develop
your application and write prototypes without worrying about boring
details like this one :D
At the end, maybe you won't need more than serialization, especially
because you plan to sell it as SaaS with one instance of gst per
customer. IMO using one image/instance of gst per customer with a
global
dump of his data is most of the time enough.
And if you want to use a Relational DB, I would use ROE with GST. I
was
amazed to discover how well integrated it is -- and this is one more
thing I should congratulate Paolo for ;-).
Cheers,
Nico (from Genova)