Home Page for the TeradataForum
 

Archives of the TeradataForum

Message Posted: Tue, 19 Oct 2004 @ 09:24:42 GMT


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


Subj:   Re: Difference between Horizontal and Vertical data redistribution
 
From:   Ramakrishna_Vedantam

Hi Pete,

Just adding to what David said.

The result rows picked up are first placed in the AMP local spool and sorted. They are then redistributed over the BYNET in a way that all values from the first sort value are placed on the first logical AMP (which is randomly selected), values from the second sort value are placed on the next physical AMP and so on, round-robin until all data is sorted. This is named as the VERTICAL distribution.

HORIZONTAL distribution takes place as data is placed into blocks which are constructed taking all values for the first sort value from the first AMP, all values for the second sort value from the next AMP and so on in a cylic way till the block is full.

Number of sessions are then used to send back the sorted blocks sequentially to the host.

There is an exhaustive usage of resources in this method. Performance enhancement is observed as there are more number of sessions and blocks are sent to the host.


Regards

Ramakrishna_Vedantam



     
  <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: 15 Jun 2023