All Forums General
pratuish 3 posts Joined 09/13
05 Apr 2016
Criteria to select Multivalue compression

Hi,
 
This may be a failrly simple question. But can someone please tell me the criteria based on which we decide the values in a column fit to be included in multivalue compression. Is there some formula or it it done based on the total values / Total rows ?
Any information in this regard will be useful.
 
Thanks

dnoeth 4628 posts Joined 11/04
05 Apr 2016

You find all details on MVC in the Database Design manual, e.g.:
http://www.info.teradata.com/htmlpubs/DB_TTU_14_10/Database_Management/B035_1094_112A/ch14.064.020.html#ww18090665

Dieter

pratuish 3 posts Joined 09/13
05 Apr 2016

Thanks dnoeth. The link is really helpful in understanding MVC.
But can you also help me to understand the values which needs to be selected.
For eg :
I have a table with total count of 1.2 billion and have 67 columns. So for me the criteria would be or exaample -
1. Select values with higher repeating values starting from 1st column.
2. Calculate the percentage by evaluation the total row count.
3. Evaluate the candidate having atlease 1% of the total row count ?
This is my evaluation but I could be wrong.
Is there a different process altogether or am I missing something here or some formula ?
Thanks

drmkd17 54 posts Joined 10/12
14 Apr 2016

I have some repeated values in my table. I should go for compression or partition for quicker access?

dnoeth 4628 posts Joined 11/04
14 Apr 2016

Compression will also result in quicker access, but mainly because the table is smaller and thus scanned faster, e.g. 30% less space results in 30% faster Full Table Scans.
But partitioning might provider n times faster access, scanning only 1 partition out of 100 is 100 times faster when they are equally sized.

Dieter

drmkd17 54 posts Joined 10/12
14 Apr 2016

Thanks  Dieter :)

You must sign in to leave a comment.