Unlock a world of possibilities! Login now and discover the exclusive benefits awaiting you.
Hi,
I had a data model consisting of the table 'Anzeigen'. This table mostly contained IDs (numeric values) e.g. SourceID or AdStatus. The description of those IDs (characters) was stored in the linked tables Source and AdStatus. There is a image of the relevant part of the data model attached.
I heard that the application performance can be improved by reducing the number of linked tables. So I created QVDs which now contain a table 'Anzeigen' with the columns SourceDescription, AreaID, AdType, AdStatusDescription, ... (hence I replaced the numeric field SourceID by the character field SourceDescription, and AdStatus by AdStatusDescription).
The result of this test was:
My question is: Was the assumption wrong, that application performance can be improved by reducing the number of linked tables?
Many thanks in advance for every comment.
Kind Regards
yes, minimum links and table reduce the time. and most of the times it depends on your table and what database you are using.
Regards,
Himanshi
Hi Nihal,
Yes, It depends on the type of schema which you used.
Hi Balraj,
thanks for the collection, but I don't really need general help regarding performance improvement. I think I already follow most of the stuff like:
- store data from sourced in QVDs and use of them in the application
- store expressions as variables
- use of variables in objects
- use of linked tables
I was just interested explicity in the difference between Snowflake and Star Schema (see post below). Nevertheless, I tried Rob Wunderlich's Document Analyzer Search Recipes | Qlikview Cookbook and it was useful. I can recommend it to anyone else!
Hi Tamil,
thanks for the overview. I can add the following:
- QVD generation for Star Schema takes longer than for the Snowflake Schema (at least in my case where I had to replace numeric values by the actual character string)
- QVW files with Storage Schema are bigger (in my case)
unfortunately I did not see any improvements. all my data is stored in QVDs
Can you post your Table view screen shot here.?
Hi Nihal,
Nihal Cakir wrote:
- QVD generation for Star Schema takes longer than for the Snowflake Schema (at least in my case where I had to replace numeric values by the actual character string)
Do you use more left Joins in your file? If yes, try to use Mapping load and it will little faster.
The yellow highlightes fields are the once that I directly added to the main tables 'Anzeigen' and 'Login' so that I could get rid of the tables Source, AdStatus, LoginStatus and PaymentStatus.