The main technique to parameterize datasources is to place your values in a location that the datasource will always look up when running. For example, you could save your parameters in a text file, or in the Text property of some metadata member. When the connector runs, it looks up that text file or member property, and uses the related values.
If you make those parameters changeable by a user through dashboards, one thing you might have to keep in mind is concurrency - what happens if someone else changes those values in the time between me setting them and me launching an import? (This could well be premature optimization: in a simple situation, with only one admin, it won't be a problem. It's only if you need your system to be scalable, that you'll have to figure out some sort of security or warning mechanism.)
