I basically use this for my packages. Something to consider is still use package configurations to denote the various environments...assuming you have a SQL Server running SSIS in DEV, QA, and PROD.

I configure an environment variable (named the same) on each server that contains the data source of the server. This requires a package configuration to be created in the package that looks at the environment variable. This way when you deploy to QA, the environment variable loads when the package runs and points to the SQL Server that contains the other QA connection strings and other config items.

This is a rough overview but do you see where it's going?