Home Page for the TeradataForum
 

Archives of the TeradataForum

Message Posted: Wed, 12 Oct 2005 @ 11:39:58 GMT


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


Subj:   Re: Redestributing large tables instead of replicating a small table
 
From:   Dunweber, Ole

Hello,

Regarding the Orange Book on PGB (Partial Group By Optimizations) - it's for R6.1.

I have taken a look into it. It appears pretty cool, but before we get to carried away note paragraph "2.6 PGB Limitations/Restrictions", quotes:

"PGB cannot be applied on PPI tables"

"PGB does not work with a join index/hash index"


Later a little bit of hope:

"The implementation restrictions of PGB will be lifted in future Teradata Database releases."

Anyway I'm encouraged by the fact, that Teradata have started to cost aggregations.

In my experience the main cost-factor when doing large aggregations is whether aggregations can be done locally or globally.

Local aggregation requires that the PI is included in the group by columns AND THE JOIN does not redistribute the big table.

Therefore I'm encouraged by this quote from the book

"Another factor of the aggregation cost is whether the SUM step is done locally or globally".

Does anyone know whether Teradata considers aggregation-cost for all aggregations, or only when PGB might apply?


Ole Dunweber
Coop Nordic



     
  <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