Skip to main content
Announcements
Qlik Connect 2024! Seize endless possibilities! LEARN MORE
cancel
Showing results for 
Search instead for 
Did you mean: 
Not applicable

QlikView Memory Config

A quick followup question regarding

I'm using QV 10 SR1.

To sum it up, QV uses lots of memory (when users are logged on or when charts are opened by the users as they are cached on the server). 

If  two users look at the same chart with the same filter, does it store x amount of memory or 2x (two users)?

Here's my server setup - duo core 7.5 GB RAM

What's the recommended numbers of users and reports (charts) for this server setting, especially RAM capacity?

Our server is currently servicing 20 users with 20+ reports (20 charts).  Should I increase memory to at least 16 GB? 

Thanks,

-Lawrence

2 Replies
Not applicable
Author

In my experience it isn't so much how many users / charts, as how much data is behind them. A table with 5 million rows & 5 columns is tiny compared to one with 5 million rows & 100 columns. So in many respect's its more a question of how big is your app?

Two things I would suggest, 1st jump up to 10 SR2 as soon as possible (lots of bug fixes), 2nd get a copy of the QV10ServerMonitor.qvw

http://community.qlik.com/qlikviews/1029#

and get it running on your server. It is very useful in measuring how both your server & your app's are holding up to your users.

In my opinion, don't skimp on the RAM, its cheap, figure out how much the box will take and thats how much you need.

Anonymous
Not applicable
Author

Hey Lawrence,

How much memory one specific document uses is a bit hard to tell without testing. If you restart the QVS service and load the document with one client you'll get a good idea of it. Then when you add additional users, expect an increase of somewhere around 5-15% of the document RAM usage per additional user.

Also, QVS should use a lot of memory. As long as you don't have any problems there's no cause for concern. QVS will cache all selection up until the Lower Working Set Limit (by default 70% of available RAM at the point of the service starting). Documents getting unloaded from RAM will bring the usage down a bit, but every selection made by a user will be cached up until this point. No point in having unused RAM when it can be used to hopefully improve performance for the next user opening the document.