An anonymous reader writes: I am currently trying to speed up some processes that slurp up and process some amounts of data from moderately sized databases of several 10's of gigabytes.
I am looking at one approach to build an object layer that will serve as a cache / mediator between the main database and the clients, but I see issues with lazy writing, data merging, and multi-user access.
I currently read data into screens from ADO recordsets, and write similarly. Therefore I am looking at some rewriting to make screens and related code run against my home-grown object layer.
What do people think of this approach, and what other approaches might be suggested? Since some screens include summary and computed information, being able to run SQL against the cache would be a benefit and save me writing some code.