We have full access to Qlikview server and publisher. We have a dataset that is too big for anyone's PC. The only way I know how to access the data and use it is to remote onto the Server and use Qlikview Desktop there.
My manager has asked if a functionality exists that lets you use the Qlikview Desktop on your PC, but utilize the resources of the server. Qlikview Desktop does provide "open in server" functionality, but it is my understanding that you cannot reload or alter the script with this approach. I need to alter the script and potentially do multiple reloads.
Thoughts? How does one interact with a large dataset on QV Server / Publisher?
Do the developing in Qlikview Desktop using a reduced dataset. For example by using the FIRST keyword to restrict the number of records loaded into a table or by adding where clauses to the load statements. After you've developed the document and are confident the front end is finished and the script to load the data is correct you can move the document to a test server and load the full data set. Then you can test further and see if reloading performs correctly and that performance of reloading is acceptable. And of course test interacting with the document too to see if everything works and performs well and looks good. If you find that things are not optimal then return to developing with the reduced dataset. Rinse and repeat as necessary,
Thanks for your reply! That is actually exactly what I have been doing! But my manager wants to know, is this the best or the correct way? Other tools let you operate on big data more seamlessly than running on a sample, putting it on the server and then reloading it.
Other tools don't load the entire set into memory. They often need a blazingly fast dbms to get acceptable performance on large data sets.
If you have a 64-bit OS on your desktop machine you can work with larger data sets if you add more RAM memory. The more RAM the larger the data sets you can work with.
Also, it's common practice to develop largely with reduced data sets regardless of the tool you use.
Gysbert Wassenaar is absolutely right, it's standard to create applications where the load is limited for development. This can also be utilized to provide a 'lite' version of the dashboard on your server.
The lite version may be just the last three days. Often users don't need to go that far back and they can use the lite version as its quick to respond. A full version holds all the data and can be slow if you are very data heavy but that's a trade off.
I pass a variable server (which can also be set in the script) and in the code interpret how much data to return.
