Home Page for the TeradataForum
 

Archives of the TeradataForum

Message Posted: Mon, 25 Apr 2011 @ 11:04:50 GMT


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


Subj:   Re: Drop ji and then recreate it to avoid overhead?
 
From:   Barrineau, Claybourne

As the other user mentioned, stats appear to be missing in the piece of the explain you posted. Stats are always a good start to addressing performance issues.

As a rule, we will not support any production processes that alter table header definitions (which you are doing when you drop and recreate JIs) unless we are 100% certain that nothing can access the respective tables while the alteration is running. Failure to guarantee this can result in user or batch jobs abending due to a 'Table has been restructured error' or (in some quirky cases) I've seen dead-locking issues that the database doesn't recognize as a dead-lock.

Granted, if you can guarantee no one will access the respective tables...great, ignore the above. If not, I think you are better off changing the respective load job to something that supports JIs. If you haven't played around with the 'Fastload to a staging / ANSI merge into a Target' approach in 12.0, give it a try. In our experience, the performance in most cases (depending on volume) is better than the traditional load utilities.


Hope this helps,

Clay



     
  <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