Page 1 of 1

Partitioning when using TeraData

Posted: Fri Aug 27, 2010 9:12 am
by Azzuri
Hi All, has anyone ever run into this problem. I have a simple job

TeraDataEnterise
|
|
Dataset----->Join----->Dataset


The join key is Key_1 and both input datasets in the join stage are partitioned and sorted by Key_1 and it's not being pertioned properly. If I create a dataset first from teradata as below it works fine

Dataset
|
|
Dataset----->Join----->Dataset


We have this issue only when reading directly from teradata enterprise.

We are running on a 2 node configuration file and using the following connection options RequestedSessions=4,SessionsPerPlayer=2

Thanks in advance

Tony

Posted: Fri Aug 27, 2010 9:15 am
by Azzuri
Sorry the top input to the Join stage is mis-alligned in my previous post.

Posted: Fri Aug 27, 2010 9:50 am
by chulett
You can fix it, you know. Wrap them in "code" tags and use Preview until everything lines up nicely.

Posted: Tue Aug 31, 2010 9:49 am
by Azzuri
Anyone have any ideas?

Could it have something to do with Teradata partioning data?

Partitioning when using TeraData

Posted: Thu Sep 02, 2010 5:36 am
by ajay.vaidyanathan
This has got nothing to do with your Teradata Partitioning. If you are using Teradata Enterprise Stage then you should concentrate on the partitioning that is happening at DataStage end and not at Teradata end. Please revisit your DataStage code with respect to partitioning and let us know.

Posted: Mon Nov 08, 2010 11:38 am
by Azzuri
Hi All, still haven't figured out this issue so created a simple job to test it. ProductCode is Varchar(100) on both inputs. Doing a simple inner join using Join Stage:

Left side of the join (Row Generator)

peekGenRows,0: ProductCode:1FL
peekGenRows,1: ProductCode:1DLCP
peekGenRows,1: ProductCode:1FKCP
peekGenRows,1: ProductCode:1FLG1

Right Side Of Join (Teradata Enterprise)

peekProdHier,0: ProductCode:1DLCP
peekProdHier,0: ProductCode:1FKCP
peekProdHier,0: ProductCode:1FLG1
peekProdHier,1: ProductCode:1FL

Both input links in the join are hashed by ProductCode and sorted in Ascending order Case Sensitive.

The result is 0 rows matched obviously as rows are going through the wrong pipe.

Configuration File :

main_program: APT configuration file: /home/dsadm/Ascential/DataStage/Configurations/default.apt
{
node "node1"
{
fastname "dcob04"
pools ""
resource disk "/home/dsadm/Ascential/DataStage/Datasets" {pools ""}
resource scratchdisk "/home/dsadm/Ascential/DataStage/Datasets/Scratch" {pools ""}
}
node "node2"
{
fastname "dcob04"
pools ""
resource disk "/home/dsadm/Ascential/DataStage/Datasets" {pools ""}
resource scratchdisk "/home/dsadm/Ascential/DataStage/Datasets/Scratch" {pools ""}
}
}

Posted: Mon Nov 08, 2010 12:23 pm
by Azzuri
Hi All, issue has been solved. Changed column extended Property to Unicode for both links and it worked fine!

Hope this helps others who might run in to this in the future.

Posted: Mon Nov 08, 2010 12:24 pm
by Azzuri
Hi All, issue has been solved. Changed column extended Property to Unicode for both links and it worked fine!

Hope this helps others who might run in to this in the future.