topic AW:Performance boost by splitting script into different tabs? in QlikView

Performance boost by splitting script into different tabs?

Fri, 14 Jan 2011 11:58:41 GMT

Hi,

I just noticed that my script was executed much faster after I split it into different tabs. What is the reason for this?

My script was very long and it takes at least 15 minutes to be loaded. As far as I can remember I just copied some lines of code from the first tab to several new tabs. I separated the load of different SQL tables into different tabs. Now it's loaded within 3-4 minutes. Is there any reason for this or is it by fluke?

Regards

AW:Performance boost by splitting script into different tabs?

rbecher — Fri, 14 Jan 2011 18:53:09 GMT

Hi,

it depends on several things.It's hard to say without knowing your script but it could be an interesting case..

Was it a SELECT with joining some tables in the first place or just a straight table? I can imaging that joining or associating on QV is much faster than on database because everything is in memory.

- Ralf

Performance boost by splitting script into different tabs?

johnw — Sat, 15 Jan 2011 02:25:14 GMT

If all you did is split your code into tabs, with no changes and no resequencing, the performance should not change because the exact same code is executing in the exact same order. So a big time difference like this suggests that you changed more than you think. But it is also possible that during the longer run, there was a lot of other work being done on the machines that had to do work to satisfy your script. There may be other explanations as well.

AW:Performance boost by splitting script into different tabs?

Sun, 16 Jan 2011 20:36:41 GMT


Ralf Becher wrote:Was it a SELECT with joining some tables in the first place or just a straight table?

Yes, I used several SQL SELECT and JOIN statements.

AW:Re: Performance boost by splitting script into different tabs?

Sun, 16 Jan 2011 20:46:35 GMT


John Witherspoon wrote:
If all you did is split your code into tabs, with no changes and no resequencing, the performance should not change because the exact same code is executing in the exact same order. So a big time difference like this suggests that you changed more than you think. But it is also possible that during the longer run, there was a lot of other work being done on the machines that had to do work to satisfy your script. There may be other explanations as well. <div></div>

That's what I thought. What I've understood so far about QlikView I would expect no difference between a single tab script and a multiple tab script. But when I saw this result I was in doubt about it. Maybe there have been some other changes to the SQL Server, I don't know. Or perhaps my former executions were at the same time than some other jobs and SQL Server had a performance peak.