topic Re: Data Size V Application Size in QlikView

Data Size V Application Size

kevinstanfield — Mon, 26 Jan 2026 18:19:17 GMT

Hi - perhaps a dumb question but I'll ask it anyway.

In terms of sizing servers based on data size (& concurrency) I'd previously used the size of the data not the application size which is considerably less. As an example for a recent customer proof of concept I have raw data of approx. 750 mb (.csv files) but the loaded application size is 57mb.

Based on what I've researched & read on the community QlikView is obviously very adept at data compression (even without QVDs) and it would seem that I should be basing the server sizing on the application size not that the data size – is that a correct assumption or is it too simplistic a view?

Thanks in advance

Kevin

Re: Data Size V Application Size

kaushiknsolanki — Mon, 29 Feb 2016 19:33:12 GMT

Hi Kevin,

Its true that QlikView compresses the data to almost 1/10 size, but you you should also know that this compression is when it stores data either in QVD or QVW, but when you open the application the RAM used by the application will be app Size * 4 (this number may increase depending on data model of application) + 10% of app size per user, to store their activity.

So you must consider this factor while deciding sizing of QlikView Server.

Regards,

Kaushik Solanki

Re: Data Size V Application Size

gregortvw — Tue, 01 Mar 2016 08:43:41 GMT

Also if the data is highly redundant, it could be more. We have an app having ~300Mb HDD size and taking ~30-36Gb of RAM while using (both, server/desktop)

Regards

Re: Data Size V Application Size

kevinstanfield — Tue, 01 Mar 2016 09:40:27 GMT

Thanks Kaushik

Re: Data Size V Application Size

oknotsen — Tue, 01 Mar 2016 11:14:48 GMT

If your question is now answered, please flag the Correct Answer.

If not, please make clear what part of your current question you still need help with .

Re: Data Size V Application Size

kevinstanfield — Tue, 01 Mar 2016 11:33:54 GMT

Thanks Grygoriy,

Could you expand on what you mean by "if the data is highly redundant"

Thanks

Kevin