Skip to main content
Announcements
Customer Spotlight: Discover what’s possible with embedded analytics Oct. 16 at 10:00 AM ET: REGISTER NOW
cancel
Showing results for 
Search instead for 
Did you mean: 
marksmunich
Creator III
Creator III

pefformance QlikiView Standalone Vs QlikView Server

Hallo qlik's,

Is there a difference in performance if a qlikview App with millions of data sets runs on a server machine with qlikview server VS the same qlikview app with millions of  data sets on a qlikview standalone desktop version with the high configuration like intel xeon CPU E52667, 3.30 GHz and 256GB RAM.  Lets us say here the Hardware for both the machines is the same.

Expert advice is appreciated.

Thanks

Marks

1 Solution

Accepted Solutions
marcus_sommer

50 M rows aren't small but very big is something different. I assume that your data-model isn't somehow suitable and you need to adjust your approaches: TableView Model - Star Or Snowflake - Does It Matter? Also make sure that there no synthetic keys or circular refrerences or calculation over loosen tables: Get started with developing qlik datamodels.

Another very important point is to reduce high cardinality fiedls The Importance Of Being Distinct, avoiding unused fields Major Update to Document Analyzer | Qlikview Cookbook and using numeric keys (maybe per autonumber).

Further the kind of calculations within the gui made a huge impact. You should avoid if-loops and use set analysis and flgas for calculations: Performance of Conditional Aggregations.

- Marcus

View solution in original post

7 Replies
marcus_sommer

In general I think there will be no real difference then the engine in both is the same. But your comparing corresponds not entirely with the reality. On your server will probably runs other tasks parallel, AFAIK (actually) you couldn't use 256 GB RAM with a laptop, the storage will be different and probably some more things will be different - very important will be the same qlikview version from server and fat-client.

I think if you want to find the differences in your case you should have a look in the load-logging and compare them. Then you might see that quite trivial things like opening the app or connecting to a database or store a qvd will be different.

- Marcus

Peter_Cammaert
Partner - Champion III
Partner - Champion III

Do you have a server as a desktop machine? In that case performance may be the same, although as Marcus pointed out, a desktop is a single user machine (probably running just QlikView Desktop) while a QlikView Server environment is not.

Servers will use a different I/O architecture and another technology to communicate with disks and RAM. There will be a difference not only with respect to hardware, OS and services, but also on the QlikView side. Running an app in QV Desktop is a bit different from opening the same document in the Access Point (different technology, network delays and such)

I think a sensible comparison will be very difficult to make...

Peter

marksmunich
Creator III
Creator III
Author

In My case, the Machine has almost server Hardware, but its running enterprise edition OS and QlikView Desktop version 64Bit edition with intel xeon CPU E52667, 3.30 GHz and 256GB RAM

marcus_sommer

I think this is a very theoretical discussion without a real added value and you will have a lot more insights if compared the log-files and not from only these application else from some more different applications.

Am I right in assuming that you are checking if you could combine qv server + qv client to avoid an increasing of your server environment or what else is the aim?

- Marcus

marksmunich
Creator III
Creator III
Author

Actually My Aim is to achieve performance for a qlikivew dashboard. the performance of the app is very slow. I have followed all the best practices and tried moving the complex  expressions from the frontend to the datamodel scripting. layer. Still it is not performing great. Hardware of the system seems also good. As mentioned the Data is very big up to 50 Million rows and so. So i am confused if the Qlikview developer edition on a desktop is failing to achive the results .

marcus_sommer

50 M rows aren't small but very big is something different. I assume that your data-model isn't somehow suitable and you need to adjust your approaches: TableView Model - Star Or Snowflake - Does It Matter? Also make sure that there no synthetic keys or circular refrerences or calculation over loosen tables: Get started with developing qlik datamodels.

Another very important point is to reduce high cardinality fiedls The Importance Of Being Distinct, avoiding unused fields Major Update to Document Analyzer | Qlikview Cookbook and using numeric keys (maybe per autonumber).

Further the kind of calculations within the gui made a huge impact. You should avoid if-loops and use set analysis and flgas for calculations: Performance of Conditional Aggregations.

- Marcus

marksmunich
Creator III
Creator III
Author

thanks a lot Marcus for your inputs.

Regards

Marks