Discussion Board for collaboration related to QlikView App Development.
I created my data model in 2nd layer.
So I don't want to call that as binary load in another qvw file rather on 2nd layer only I will create dashboard..
If any suggestions wts the advantage if I use it as binary and create only front end in different dashboard.
Please exprets can comment bcs, I felt why to unnecessarily increase one more layer in the name of 3teir architecture..
IF really needed whats the advantage (anything like dashboard on 2nd layer may b 200 mb and if used binary might become 100mb )..
Request you to comment with valid proofs why Binary wt it ll do?
why should I create 1more layer? which will increase my task running time by adding 1more application?
Why Qlik recommend 3layer..(nt looking fr docs as I have many but need justifications)
Else I think in 2nd layr we can build dashboard..Plss comment as its going to effect entire my organization??
Agreed but imagine if your showing aggr data in the 3 layer UI and you need to had one more stream of data then ? you no need to touch the front end ..Just need to change the 2 layer and rest remains the same. and same thing goes with the business logic also ...
You could do it no issue with that ...it depends on the situation..
*one thing I found is maintenance and handling the complex logic and data
*Re-use of QVD's formed in 2 layer(Aggr ed ) can used in other dashboard were its required
I found helffull ,only when reusability of datamodel comes in to picture,else no speed nthn..
Bcs handling logic can be done in same file..
Any one read about this frm Qliktech stuff?
Maybe it's helpful to take a look on the main-idea behind the use of numerous layers which is quite simple. It is to break a big task down into several smaller ones and then splitting them again into more and more simpler ones - to be able to distribute them between various resources and to standardize them.
Of course this approach caused more working steps and adds some degree of overhead to administer them but in the end it will run faster and be more stable as with the big task. And this isn't an idea of qlik or from the IT area it's valid for the most parts of the daily life ...
If you run a rather small qlik-environment it's not essential to use a 3-tier (or more) architecture but if it's growing and growing you will get more and more benefits by splitting and standardize your tasks. This meant if you have just a few applications you don't need to create a big (overdesigned) environment but just keep it in mind that it made sense to adjust it if it grows.
Thanks for response.
I got ur points,its kind of common sense wthr to go fr 2 or 3 for well organized structure.
U mensioned,it will run faster,But how it will run faster?? As I will be having one extra layer(Its like running 2nd layer and again running 2nd layer in 3rd layr bcs with binary QV has to create Data model internally frm 2nd layer)..
IF we go with that,Dont you think we r loading data model twice unnecessarily which is burdening my server and performance.
So MY question again boil down to :
whether Tht Binary statement doing something thr?? (Like handling in 0's or 1's or reducing data modelsize??)
Or its just a statement to load 1 QVW file into other??
I must admit that I never measured this directly. But if I look on my automated daily running tasks (only for updating applications and without any export/printing tasks) the datamodels = layer 2 takes nearly 45 minutes for the update - the reports to them = layer 3 are updating per binary load in about 4 minutes. This creates an overhead from about 10% - not very much to keep different things separate.
But this isn't the whole truth - some of the reports share the same datamodel and after the binary load I adjust them by adding some small data (for example fields for dimensions/expressions for interactive charts - which are not needed in each report) and/or dropping some tables and fields. Also the available variables, dimension-groups and the set access rights per section access are partly different. I think if I would create for each of them an own combined datamodel-report my overall run-times would be longer as with my now use 3-tier architecture.
I believe there are also some more practically reasons why a multi-layer architecture might be sensible. Because of my growing qlik-environment I will switch in the summer (new business release + migrating to QV 12) to a 4-layer architecture with a further splitting of generator- and datamodel-tasks and I'm sure that I will benefit from them.
We can reuse the data model and can integrate more additional data sources :
In my recent dashboard build followed 3 - tier:
1. Reused existing complex dashboard datamart layer qvw(It was an old app which was build some 2 year's back)
2. additional data sources were integrated and then created one new set of dashboard
3. After all these we have one new set of dashboard with more data, new security and much more added advantages and more over it was specific to our own set of business people.
HI,Thanks fr response
I boiled down my question as above..Plss say, if u have seen anything of tht sort..Bcs wts been going on in thread I have also done and experienced.