Home Page for the TeradataForum
 

Archives of the TeradataForum

Message Posted: Mon, 24 Apr 2006 @ 21:40:24 GMT


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


Subj:   Re: Performance issue in Mload (with INMOD)
 
From:   Diehl, Robert

Rishi,

Mload Session error - Must see the error code. Could be an all circuits busy. Do you have a ton of Multiloads that all start at the same time. This could be because you had fewer Parsing Engines thus fewer sessions (120 per PE) available and ran out of sessions. You want the number of sessions defined anyhow as too much overhead is used when logging on many sessions. Also MLOAD doesn't necessariy perform best with more sessions. You can do this with a mloadcfg.dat file. This file can contain min and max sessions for multiload See MLOAD documentation for specifics. However, I suggest you dynamically add the sessions= to your mload before running. This can be done at size level or individual job level. This way you can easily tune every job and also make the change in as few places in a system upgrade.

However, if you are using a ETL tool it may not allow you to use dynamic substitution.

For your performance problem, make sure there is not a lot of skew in your target table (make sure the values in PI are unique or nearly unique).

Distribution of data is different on the new configuration which may exacerbate a poorly chosen PI.

I have a feeling the MLOAD is waiting for data.

Make sure your Reference id lookup in the inMOD are using PI.

Also make sure you are not being blocked on these lookups.

Also check on Secondary index or Join index problems that might be impacted by the maintenance.

Have you thought of just doing a straight MLOAD to a stage table and then applying the Reference table lookup in Set logic? I bet it would be much faster


Thanks,

Bob Diehl
Travelocity.com



     
  <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