Home Page for the TeradataForum

Archives of the TeradataForum

Message Posted: Mon, 22 Oct 2007 @ 17:32:42 GMT

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

Subj:   Re: Compression Algorithm
From:   Dieter Noeth

Barner Eric wrote:

  Can you expand on your below comment.  

  "Now there are 64-bit systems, so this is even worse, due to pad byte fields and alignment issues :-("  

It's more complicated to calculate if an additional presence byte will result in an actual row size increase. And if converting a Varchar to a compressed Char will result in a different size for the Varchar Offset Array.

Check the Database Design manual
Chapter 15: Database-Level Capacity Planning Considerations
Base Table Row Format

  This is a harrowing thought when thinking about converting from our 32 bit MP-RAS to 64bit Linux.  

  I have not put that much thought to the idea, but from what I can understand from your statement, this could have a significant performance impact in migrating to a new OS.  

At least perm space usage will be increased. Quote: "but studies with more realistic customer data in the field indicate that the identical row on a 64-bit system is likely, on the average, to see a size increase closer to 7 percent over its size on a 32-bit system"

  I guess my question is, in your opinion, will the switch to 64 bit > > significantly degrade the effectiveness of MultiValue Compression?  

Probably not, it's just much harder to pre-calculate the actual benefit.


  <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