Home Page for the TeradataForum
 

Archives of the TeradataForum

Message Posted: Sat, 04 Jun 2005 @ 19:59:14 GMT


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


Subj:   Re: PPI table - empty partitions.
 
From:   Dieter Noeth

Robert Diehl wrote:

  2) using Sampled stats on PPI column request_dt caused much the same situation as in case 1. The data volume was fairly uniform across all partitions.  


Full ACK.

The word "sample" is missleading, it's not similar to a "select .... sample x".

It's simply scanning the table on each AMP, i.e. the first datablocks per table are read.

Unfortunately a partitioned table is sorted by [partitionnumber, rowid]: If there are 100 partitions a scan of 2% will result in 2 or 3 distinct values and the remaining 97 will be unknown to the optimizer.

I've never seen a warning about that "feature" within the manuals, would be a good candidate for an error message.


Dieter



     
  <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