More advanced topics of qlik datamodels

    Beyond the basics of data modelling and the advanced topics of qlik datamodels there are more things which you might need if the amounts of data are increasing or if your datamodel is more complex. This meant if you have to combine multiple data-sources within more than one application and/or various applications are sharing a datamodel or it's even a cascade from (more or less nested) datamodels.

     

    The view of things changed if you haven't only a few applications but an entire reporting environment with dozens or hundreds of applications and there are probably more then one people which work and maintain on it at the same time.

     

    At first you should begin with a systematically approach. My suggestion is to start with a white paper and write down the requirements and how could be they solved, read books and postings like this, ask people which have done a similar work before and also consider to use external help for the difficult things - and be sure you have a plan how the end-result should look like. Nobody builds a tower without a plan from the architect.

     

    But of course no plan is perfect and covered everything, you will need to adjust it several times and you will need by some aspects many attempts to gain your aims - but you shouldn't have a simple trial and error approach.

     

    Therefore: Here a few links about the most importance things like: data-architecture (layer), folder-structure, naming standards, star-scheme vs. snowflake-scheme, various methods to match data (joins, mapping, concatenate, table-associations), security-topics and many more.

     

    Best Practices for Data Modelling

    Data Modeling issues and solutions

    The Optimal QlikView Data Structure

    Best Practices in Data Modeling

    Three Tier Architecture

    QlikView Data Architectures

     

    Beside those more theoretical deliberations exists an own group within the qlik community: QlikView Deployment Framework. to make such things easier and more practically.

     

    Main topic here is the development of datamodels within qlikview but those are always imbedded in bigger enterprise structures therefore you should also think a little bit outside the box then you won't be successful with an isolated qlikview system.

     

    qlik-view-project-methodology-handbook-v-10

     

    Another important point is security and how you could manage Authentication and Authorization. The used authentication will be mostly defined on company level and could be often directly imbedded in qlikview with SSO. Authorizations instead are mostly applied on application level with section access (by use of the publisher loop and reduce is possible, too).

     

    Beside those rather bigger things there are of course further topics on how qlikview worked under the hood which you should to know to improve the performance and the maintainability:

     

    Symbol Tables and Bit-Stuffed Pointers

    Symbol Tables and Bit-Stuffed Pointers - a deeper look behind the scenes

    ByteSizes from Values in QlikView

    The Importance Of Being Distinct

    A Myth about the Number of Hops

    Data Modelling: Clarity vs. Speed

     

    Another interesting point to the above links are synthetic keys - my personal recommendation is you should avoid synthetic keys but of course each coin has two sides and synthetic keys aren't a general problem. If you want to dive (really) deeper you should read this: Should We Stop Worrying and Love the Synthetic Key?

     

    There is some content-overlapping within the above used links but I think the complement to each other is useful and of course you will find many more interesting postings here within the qlik community to these topic - the notes here are a good starting point to go further.

     

    Have fun!

     

    Marcus Sommer