I'm not sure on the average size of the tables really. Currently the etl is a c# app that exports some tables whole and some tables just new entries based on datetime stamps.

If we went for exporting say 12 tables whole every time then maybe 200mb's as often as possible.

I guess I'll have to look into the powershell, ssis options.

I could maybe get the c# app to do a backup/restore (to a different name) of the database/certain tables then run some syncframework code with azure. Then when complete start the process again.