Home Page for the TeradataForum
 

Archives of the TeradataForum

Message Posted: Tue, 30 Nov 2004 @ 09:57:00 GMT


     
  <Prev Next>  
<<First
<Prev
Next> Last>>  


Subj:   Physical design/implementation for market basket data
 
From:   Ole Dunweber

Hello,

This is mainly for retailers, but I encourage anyone to answer. We are about to reengineer our implementation for market basket data. The current design uses a denormalized basket detail table (hence with a lot of redundancy).

PI is basket_id. So the PI has been chosen to optimise join access, for the very important affinity analyses (equi-join on basket_id).

The questions are:

1) Does anyone use a normalized model - that is a basket_header table and a basket_detail table?

2) If so, what is the PI for the header and detail table?

3) Does anyone use PPI for this/these table? (If normalized we could choose to include transaction_date on the detail-table, using this as partition-column)

4) Has anyone succeeded in choosing a PI suited for star-joins =96 e.g. (product_code,transaction_date)? I guess that would have to be complemented by a SJI with PI= basket_id, to help affinity analysis


Best Regards,

Ole Dunweber



     
  <Prev Next>  
<<First
<Prev
Next> Last>>  
 
 
 
 
 
 
 
 
  
  Top Home Privacy Feedback  
 
 
Copyright for the TeradataForum (TDATA-L), Manta BlueSky    
Copyright 2016 - All Rights Reserved    
Last Modified: 27 Dec 2016