QlikView server processing of client requests and timeouts

[reanfadyl]

‌Hi Experts,

Can someone shed some light on what order QlikView Server/web server processes end user selections?

if one user asks for a table chart that has say 600k individually calculated data points (i know this is bad practices but this is just theoretical), and then a second user asks for for a line chart that has say 16 calculated data points, will the second user have to wait for the first user to get their answer?

Will the second user potentially timeout before they get their result?

Does the number of CPU cores available make a difference?

Thanks in advance

Rean

ps I have read the scalability test white papers etc.

[swuehl]

If I read this correctly

Logical Inference and Aggregations

the calculations should be asynchronous and multithreaded. So "600k" user should not block "16" user.

Probably hic‌ can explain this much better.

Regards,

Stefan

[reanfadyl]

‌Thanks I took a look at HIC's post but it is not entirely clear if it is asynchronous across multiple users e.g. but  yes The  aggregations for each visible object for one user are definitley async

[hic]

All requests are asynchronous and multithreaded.

A heavy request will take most of the processing power also across multiple cores, so that other heavy requests will have to wait. But there is also a function that should prevent a heavy calculation from taking all the processing power, so there will always remain some capacity for smaller requests. Like a fast queue in the supermarket.

HIC

[swuehl · Accepted solution]

Thanks, Henric.

Rean,

and to answer your last question, adding more ressources to share between users should definitely help

(well, you already know the scalability white papers).

[reanfadyl]

Thanks to both of you for your help.  Time to get on with getting another supermarket lane set up I think.