Skip to main content
Announcements
Introducing Qlik Answers: A plug-and-play, Generative AI powered RAG solution. READ ALL ABOUT IT!
cancel
Showing results for 
Search instead for 
Did you mean: 
manojkulkarni
Partner - Specialist II
Partner - Specialist II

Suggestion on Data Model

HI Friends,

I have multiple fact tables having date & Key field are common between them. Each table have 20+ columns with 10M records.

What is the best approach to handle ? I should be able to see data for any slicing and to a lowest granularity.

Please suggest

7 Replies
marcus_sommer

It's not enough information to give an advice. Did your application work in general? How is the performance?

- Marcus

manojkulkarni
Partner - Specialist II
Partner - Specialist II
Author

HI Marcus

Still I haven't decided on Data Model. I am in initial stage of Data Model, so would like to get suggestion on how to handle multiple fact table ?  Which one is better  Concatenate or Linked Table when we have around 10M records per table..

marcus_sommer

I would rather try to reduce the number of fact-tables by joining, mapping and/or concatenate but it will depend on the data within your tables which method could be applied for which part of your tables. I assume you will need all of the above mentioned methods and if it's not possible/practicable to reduce it to one single table you will need a link-table between them - this meant a link-table between two large fact-tables is better then a or several link-table(s) between multiple fact-tables.

- Marcus

karthikeyan1504
Creator III
Creator III

Hi Manoj,

The three tier architecture can be followed here as explained below.,

1. Extract the multiple fact tables and load it in separate qvds.

2. Concatenate the tables to make one fact table and store the second level of qvd which will act as one fact table.

3. Use the second level qvd for the dashboard.

Hope this helps!!!

Warm Regards,

Karthikeyan.

jagan
Luminary Alumni
Luminary Alumni

Hi,

If most of the fields are same then it is better to go with Concatenate with an additional flag to identify the datasource, if a less number of columns are same then go for Link table.

Regards,

jagan.