topic Re: Mashup Initial Load Time in Integration, Extension & APIs

Mashup Initial Load Time

pnowicki — Fri, 19 Aug 2016 08:59:21 GMT

I have a single page mashup in which I only load the objects when needed. On the initial load, I only load 5 objects (including the current selections box). The qlik sense document is 20mb. However, the initial load takes about 15-20 seconds to render the first objects. After the initial load, the entire mashup runs very quickly including selections. I do not use angular framework but I do use dynamic loading feature where I only call the objects I need and I also destroy objects after changing between hashtag divs.

Given all of the above, is there anyway to decrease the initial load time? For example, is it a potential hardware issue on the qlik sense server? I'm running on a virtual machine with 64gb ram. My users haven't complained yet but I find it very annoying.

Re: Mashup Initial Load Time

ErikWetterberg — Mon, 22 Aug 2016 09:00:26 GMT

Hi Paul,

Not so easy to answer.. You will need to make a bit more investigation to find out why it's slow. Stuff you should check out:

Does it take the same time if you make sure the app is loaded in server memory? For example if you open the mashup again a second time, does it take just as long?
Is it a specific object that is slow? Try commenting out all objects, then add them one by one and see what happens
use the debugger and check what happens on the web socket.

Hope this helps

Erik

Re: Mashup Initial Load Time

pnowicki — Mon, 22 Aug 2016 11:17:30 GMT

Erik,

Thanks for the reply. I've been able to speed up the load time based on your steps. Here is what I did:

1. I checked the load time first time versus after it was in server memory. It was faster after the server had opened the file.

2. I found that I had a few objects in the application that I wasn't using in the mashup. I made sure only objects displayed in the mashup were left in the qlik sense application and updated the javascript to eliminate any open objects calls. This clean up seemed to help the most.

Bottom line: Making the qlik sense application leaner sped up the mashup. Originally, I thought only the type of objects and number of objects I called from the javascript would change the performance, but I guess it is obvious that the qlik sense app itself should also be optimized.