One Simple Trick That Will Significantly Boost Your QlikView Performance

By Barry Harmsen

Edit April 2nd 2015: Yes, of course this is an April Fools joke, hope I made you smile. The Masters Summit is no joke though, and neither is the $300 early bird discount that you will receive until April 15th 2015. Want to really take your skills to the next level? Join us in San Francisco on May 4 – 6, or New York City or Copenhagen later this year.

Amazing Business DiscoveriesToday I am going to let you in on a big secret.

People often ask me how I always get great performance and fast response times out of my QlikView applications. I usually put on a very serious face and reply that this is the result of years of experience. Experience gained by doing hundreds of implementations, practice, experiment, study and continually refining my skills and knowledge.

You won’t believe how hard it is to not burst out laughing when I tell this to people. The fact of the matter is that it is all a gimmick!

There is only one simple trick that you need to know to instantly boost your QlikView performance. This is the secret that Qlik and the Qlik Illuminati don’t want you to know, but today I will reveal it all!

Are you ready for the truth? Here it comes!

Keep the lines short and straight

 

There, I said it. Keeping the lines in your data model as short and straight as possible will instantly boost the performance of your QlikView application. Consider the data model below, it has very long lines between the tables and none of them are straight:

This data model was laid out use the Auto-Layout function, clear proof that Qlik wants you to build suboptimal data models so they and their partners can sell you expensive performance tuning services!

This data model was laid out use the Auto-Layout function, clear proof that Qlik wants you to build suboptimal data models so they and their partners can sell you expensive performance tuning services!

 

“Why is this important?” you may ask. Well, every time you make a selection in QlikView, the logical inference engine automatically associates the relevant data within the entire data model.

The longer the lines between tables are, the longer it will take the data to travel through that line.

Furthermore:

Sharp angles in the lines between tables also decrease performance, as the data will have to slow down to travel around corners.

 

Now consider the alternative data model below. The lines are short and straight. In the case of this model,

Making the lines 50 times shorter resulted in 5000% better performance!

Wow! What an Amazing Data Model!

 

But wait, there’s more!

 

Gravity Data ModelingProtip: Gravity is your friend

Some times it is not possible to have completely straight lines. For example when there are many tables in your data model. In that case, placing your dimension tables below your fact tables will add a gravity boost, the data will effectively fall down through the line.

 

Protip 2: Make the lines wider for even better performance

While you could have figured out the previous tips by experimenting, there is one final performance tweak that, up to now, has only been made available to the most elite of QlikView Partners. The so-called fat pipe mod makes the lines much wider, greatly increasing the amount of data that can ‘flow’ through the line at the same time.

Fat lines

The ‘fat pipe mod’ enabled on a data model. Notice how much wider the lines are.

Want to apply this amazing mod to your own QlikView installation? Scroll down for more information…

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

… almost there …

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

It's April, Fools!

 

Hopefully you had already guessed this before you scrolled this far, but of course there is no silver bullet when it comes to QlikView performance. However there are many tips, tricks, techniques and methods that combined can significantly improve the quality and performance of your QlikView applications and increase your effectiveness as a developer.


Masters Summit for QlikView - San Francisco

One place where you can learn about these tips, tricks, techniques and methods is at the Masters Summit for QlikView. This 3 day training event is designed to take your skills to the next level with sessions about Data Modeling, Scripting, Advanced Expressions & Aggregations, Data Visualization and much more.

The next edition of the Masters Summit is held from May 4 to 6 in San Francisco. Besides the core sessions, new sessions focusing on Qlik Sense for QlikView Customers and NPrinting have been added to this edition.

Even better, if you register before April 15th you will receive a $300 early bird discount, and that is no joke…

About The Author

Barry Harmsen

Hi there, I'm Barry and I'm a Business Intelligence Consultant at Bitmetric and based in the Netherlands. Originally from a background of 'traditional' Data Warehousing, Business Intelligence and Performance Management, for the past few years I have been specializing in QlikView and a more user-centric form of BI. I have done numerous QlikView implementations in many different roles and industries. In 2012 I co-authored the book QlikView 11 for Developers. You can follow me on Twitter at @meneerharmsen.

14 Comments

  • 1
    johan
    April 1, 2015 - 11:32 | Permalink

    Strange, my performance is always better when i place the dimension tables above the fact table.

  • 2
    April 1, 2015 - 12:42 | Permalink

    This method for shortening the lines between link and dimension tables is something that I was using for decades. But nowadays I use the join function to join all these tables together. No more lines. And performance is really dramatically, significantly, too obvious, insanely and extraordinairy improved by that method.

  • 3
    April 1, 2015 - 15:00 | Permalink

    I get the best performance when I’m using 10% zoom and source table view.

    • 4
      April 1, 2015 - 17:48 | Permalink

      An added benefit of using 10% zoom is that it makes your data 10 times smaller!

  • 5
    Aran
    April 1, 2015 - 15:26 | Permalink

    Is the gravity effect augmented on high gravity planets? We are thinking about re-locating our servers to Jupiter…

    • 6
      April 1, 2015 - 18:01 | Permalink

      It might, but the network latency would probably negate any benefits of relocating to high gravity planets. One suggestion from the Scalability & Performance Center is to place your servers in a centrifuge. This gives you the benefit of high (artificial) gravity while maintaining low network latency,

  • 7
    Rob Wunderlich
    April 1, 2015 - 19:33 | Permalink

    Like Seebach, I use the Source Table view to eliminate synthetic keys. It’s so much easier and faster than scripting and reloading.

  • 8
    Andrew McIlwrick
    April 1, 2015 - 22:31 | Permalink

    File -> Reduce All works for me !

  • 9
    Roberto Belmonte
    April 1, 2015 - 22:44 | Permalink

    🙂

  • 10
    April 2, 2015 - 10:58 | Permalink

    How this data gravity concept is effecting cloud deployments? I mean it should clearly be related to the cloud altitude and rack level..

  • 11
    Ellen Blackwell
    April 3, 2015 - 21:01 | Permalink

    Hysterical!

  • 12
    Andrew
    April 12, 2015 - 01:43 | Permalink

    Hello QV crew
    ** pokes head in***
    Just completed developers course and think my head is still spinning. Learned enough to realise QV is a super exciting data analysis world….
    Tips on fastest Ways to get up running?

  • 13
    MARK
    February 3, 2016 - 20:12 | Permalink

    yes, this was funny at the time but now its turning up in google. respectfully request that page be be removed or tag it such that it will not be near the top of the results.

  • 14
    Thomas
    December 22, 2016 - 22:37 | Permalink

    I would agree. It’s annoying that this shows up in google searches

  • Leave a Reply

    Your email address will not be published. Required fields are marked *

    Read previous post:
    Target lines on a bar chart using error bars

    During my internship at Bitmetric one of my project goals was to build a QlikView application to benchmark logistical companies....

    Close