So, for example, if you had an extension of a pie chart (for some reason), you could have that pie chart based on a billion rows of data, but all you really need to display the chart is the size of the pie pieces. So the calculating of those billion rows can be done on the QlikView side, and the values of the slices can be passed into the extension. If you have to have larger buts of data, you can also work around in other ways by using pagination in the extension object where you only load a certain amount of data at a time. So if you had to have a giant table of data, you could potentially just show the first 20 rows then have the user page through the rest.
Does that make sense? To sum up, yes, extensions work on a client side technology which would make heavy lifting a bad idea inside the extension, BUT, it's not difficult to use QlikView to do the heavy lifting BEFORE sending the data into the extension.
Thanks Brian sharing the insight. So we will need to do the heavy lifting on our data model side so that all the calculations are done there.
Just one thing and it may be silly question to ask. Can we do anything here with server side java scripting so that it executes on the QV rather than on client?