It is important I explain how an ETL approach was implemented in projects before Qlik Sense came into play as they will be different and should be planned differently too.

In an IT project, it is very common to have at least three different servers, one for development, one for testing purposes or user acceptance (UAT), and one production server. You may call them different things or might have an extra server, such us preproduction. What matters here are the concepts. The structure might change slightly.

In a conventional project, you would develop in your development server. Your data sources would point to the development server. Most of the time, you wouldn't have access to live data or data source from a development environment. 

Once your development is done, you would deploy your files to a UAT server where testers would try to crash what you built and find issues. So far, the development and UAT servers are in different infrastructures, and they may point to different data sources.

Once everything is approved in UAT, you would deploy your project on your production server. Your application would point to a database with live data, and it is here the end users will access the application.

The following screenshot shows what the approach to an old ETL architecture would look like:

This screenshot represents the standard waterfall ETL approach: everything protected and divided into compartments. Now, take all this and throw it in the bin as things have changed considerably with Qlik Sense.