I have 3GB QVW files. I run it on a mahine with 160GB memory 。 But I find it cost too much time and alway occur no response in windows developing environment when I open some tables. It is the same in website . Whether is QV unsuitable for such large data ?
Will it be possible to break this application in sub parts and then use like break the model for different sub application. By doing this you data model get decrease and take less space but you have to maintain different application. Why not you try the Binary load of the applications from the main application and in binary application drop the tables which is not necessary and design the reports based on your reports.
There are several approaches to improve performance of your QVW’s. Based on the information you’ve given it’s hard to give you a good advice. So here are some general points you should consider:
Is it possible to reduce the amount of data loaded into the QVW? For example drop unused fields and tables, narrow the date scope of your QVW
Have you practised the principles of good scripting > is your data model efficient in hosting the 3Gb of data?
Do you use complex expressions and calculated dimensions in your front end. These are performance killers and should be simplified by preparing them in the script.
Think about scaling out your server to a clustered server
Maybe some hardware tweaks are needed, like turning NUMA off
I’ve attached a document that can help you out regarding the last two points.
With a 3Gb QVW and 100 users logging on to it I suggest you scale out to a clustered environment. For example two QlikView Servers with both 16 cores and 128Gb. For this you'll need two QVS Enterprise Edition licenses.
Also clustering the webserver might be a good idea to get a better uptime of your accesspoint and improve the response.
