Teradata enterprise Vs Multiload stage

ag_ram · Post by **ag_ram** » Mon Apr 06, 2009 2:07 am

Hi,

Datastage version: 8.0.1

I have a requirement to export ~500 million records from teradata. There are two tables that I need to join to extract the data. The source tables are partitioned.

Could you pls let me know if its better to use Teradata EE stage against MultiLoad stage.

The documentation and some of the forum questions suggests both these stages when used as the Source Stage will internally use the FastExport utility.

How different is the Teradata connector stage when compared to the above two since connector stage also uses FastExport internally. IBM suggests Connector stages to be the future. Does this mean EE and Multiload stages could be deprecated in future.

Thanks

hamzaqk · Post by **hamzaqk** » Mon Apr 06, 2009 5:33 am

Connector stage is actually the Teradata TPT API, which means although the protocols being used behind are fast export and fastload but it will be much faster than Multiload and Fast export itself as it creates multiple instances and then process the data (not to be confused with sessions).

If the join query is complex i would suggest that you write a BTEQ and call it in the command stage. 500 million records is a bit too much for DS to process as far as my experience is (Especially for server edition). TD will always process the records faster provided you collect proper stats on the joining tables etc.