I believe this might be a broader discussion, but we are trying to move our entire stack from QlikView to QlikSense.
How should we approach this problem? I believe we need to define how the things should function differently here and cannot just replicate what's available in QlikView. Are there any best practices or case studies around this?
Below are some implementation challenges we are facing/evaluating:
1. Need to create extra applications/sheets to simplify expressions, to enable easier self Service
Question: We are creating copy of applications with different business rules. In QlikView we were handling them in long/complex expressions. What is the best practice with QlikSense? Will this not increase maintenance?
2. Need of a lot of extensions in QlikSense - for e.g. Single Select in ListBox, Top N user input, Time Period Selection/Action on buttons, Container, etc.
Question: Will these extensions cause any performance problems, since we are dealing with high data volumes (300-500 million rows), if yes then are what are the alternatives, and what type of extensions cause major problems?
3. Need to create lots of extra Sheets due to space constraint
Question: Charts are now getting spread across multiple sheets, not allowing user to see the entire picture in a single frame. Should we think of reducing the number of charts and allowing users to create more analysis, or maybe add some charts to master data elements and allow users to drop those charts from there? What will be recommended here?
4. Implementing Data masking for a specific user group (using expressions) is challenging, since users have self service access. Should we further create copies of the App for this use case?
Please feel free to add any more challenges you have faced during this exercise, and what worked well for you.
The Qlik Sense front end is different to the QlikView front end, as you allude.
The one general thing I would suggest is to work with the Qlik Sense front end and not against it. Attempting to exactly replicate what you have in QlikView into Qlik Sense could well end up painful, time consuming and substandard. If you do exactly replicate what you had in QlikView into Qlik Sense then one could ask the question as to why bother with all the effort to just end up with what you had.
Where possible initially do everything Out of the Box.
That will probably not be achievable and extensions are one of the strengths of Qlik Sense. Create a defined collection of good extensions that meet your needs, but make sure you have the skills needed to create and maintain these extensions either in house or external resources. Open source extensions can be downloaded from Branch / GitHub, but - I repeat - do make sure you have the skills needed to maintain them. They may not exactly meet your needs and you will need to be able to support them over Qlik Sense upgrades especially if any API's get deprecated.
I am an advocate of an Agile approach as opposed to Waterfall. Start small and accept that your first migrated app will not be perfect and nobody, especially non techie users, like change.
Q. How many folk singers does it take to change a light bulb ?
A. Twelve. One to change it and eleven to write a song about how much they loved the old one.
Agile allows one to fail fast, fail often and then succeed quickly. Waterfall allows one to fail slowly, fail once and realise you are in a cul de sac. Other viewpoints are available from Waterfall fans.