how to improve Nprint performance case study

[ronaldwang]

Hi, I need to schedule to run around 20 Nprint reports from my computer and my initial aim is to produce each report in 2 mins so i monitor the performance while i am building the QVW and Nprint files. up until earlier version i am still ok with the performance, the initial time to complete the first report is about 1 minutes but as i add on more content to the QVW files and Nprint files, the performance drop to around 3 minutes per report. but then after i make a few small changes (mainly as requested to split a few pages to more pages) then average time to run a report is about 9 mins which is not really ideal. thus i want to see how can the performance could be improved:

Fact:

Processor: Intel(R) Core(TM) i7 -4600 CPU@2.1 GHz

RAM:8 GB (7.71 GB usable)

QVW file size :480MB (initial 330MB)

Number of objects:2400

Number of sheets:21

Excel template size:100KB

most of the information is exported to excel by straight table or cells without formatting.

Things have been done:

I have gone through all objects and remove all unnecessary objects.

i have reduced the template to be used to the minimal：6

Constraint: the data model should be pretty optimal and would not be changed.

What i should change to improve performance?

thanks

[Lech_Miszkiewicz · Accepted solution]

Welcome back Ronald

You are not using dedicated for NPrinting document, but a document which is also accessed by people, right? I would suggest creating dedicated document in order to minimise No of columns and rows in your qvw file so it only contains objects and fields and data used for NPrinting.

see my comments here:

QVW size limitations with NPrinting 17

other thing - do you use any images from QlikView? - for example if you are producing charts as an image the same can be done much quicker by embeding XLS charts based on Qlik data. Exporting data rather than an image is much faster i noticed.

cheers

Lech

---------------------------------------------

When applicable please mark the appropriate replies as CORRECT. This will help community members and Qlik Employees know which discussions have already been addressed and have a possible known solution. Please mark threads as HELPFUL if the provided solution is helpful to the problem, but does not necessarily solve the indicated problem. You can mark multiple threads as HELPFUL if you feel additional info is useful to others.

[Ruggero_Piccoli]

Hi,

I agree with Lech and I would to suggest to give a check also to this conversation. Lech's suggestion is duplicated there 🙂 but there are other suggestions that you can implement.

Cheers,

Ruggero

[ronaldwang]

Thanks Lech, you have always been very helpful

[Lech_Miszkiewicz]

Hi Ronald - does it actually help to solve your issue?

regarding creating a version only for NPrinting i usually suggest deleting all user sheets and leaving only sheets used for NPrinting, then i suggest running Rob's DocumentAnalyzer (Document Analyzer V3.0 Update Available) in order to checks which fields can be dropped.

cheers

Lech

[ronaldwang]

Hi Lech,

I will try and let you know and I think they will to some extend improve the performance. and I am not exporting any chart to my report and have removed all charts from my app, but will try Rob's DocumentAnalyzer to check which fileds can be dropped.

But I have concerns though but dropping fields: as I also use cell and columns, but dropping the fields, how will that affect the current reference in the existing templates?

I also notice that the performance drop when my file increase from 330m to 480mb, which is due to the data size in crease. as by removing a few objects I could not see much change to the size of the files.

[Lech_Miszkiewicz]

Hi Ronald,

When talking about fields to be dropped i meant FIELDS FROM YOUR DATA MODEL and not fields from your charts. You have to understand that QlikView document has a data layer, script layer and visualisation layer. I am under impression that you so far have taken care only of visualisation layer!

Go to script layer now and remove from there all fields which are not used in frontend in NPrinting app! You have to strip your whole QlikView application from not needed data. Examples:

• you are saying data is growing - how many years of data are you holding there? are you using all those years in your NPrinting report? Maybe you can load only last month, or last year of data only?
• in terms of fields - imagine that you have Customer Table with following fields and in your NPrinting application you are only reporting sales by Customer name then in your QlikView script you can apply drop field statement for fields marked in bold/underline:
  • Id
  • Name
  • Number
  • Address 1
  • Address 2
  • Address 3
  • State
  • Country
  • Phone
  • ContactPerson1
  • ContactPerson2
  • EMail
  • Longitude
  • Latitude

The biggest benefit you will get when you drop fields from your fact table.

BTW - can you go to your qvw application and open it in QlikView desktop and screenshot a Table Viewer layout where you will see all tables and all fields?

cheers

Lech

[ronaldwang]

Hi Lech,

you are right, that is why I mention in my question that the constraint is that  I won't change the data model as it is produced by another team.

[Lech_Miszkiewicz]

Well - in this case you are left emptyhanded - the main goal is to change datamodel so it responds quicker. As performance of NPrinting relies on QlikView performance changing data model and removing columns from there is my starting point!

You do not have to change the logic in your model, you can simply apply "DROP FIELDS" statement at the end of the script which just drops unused fields.

You can improve performance very very significantly!

cheers

Lech

[ronaldwang]

Thanks Lech