I am working with Teradata DB and some programs in SAS language. I think that when you use a PROC SQL to do a query, SAS get the info from Teradata, load all the data in SAS tables, solve the query and then delete the temporary tables instead of solving the query in Teradata DB, am I right?
I have strong views on this subject and it often brings out the worst in me... So apologies in advanced :)
In short, no something is wrong.
- more comments -
Do you actually *want* to create a copy of the data? I expect not.
I use Teradata as the sole source of my data mining analysis, and everything is processed on the Teradata box. I use a competitor (to SAS) data mining tool which converts my data analysis into SQL and the Teradata box handles this SQL as it would any other query. Admittedly the SQL that gets autogenerated is not formatted at all well and reads really badly due to complex nested statements often needed in my analysis (and the SQL code is 'hidden'), but it is optimised and runs blisteringly fast on the Teradata box.
I don't use SAS, but the intention of their partnership with Teradata is to enable more complex analysis to be conducted . I have read that there exsit or planned UDF (user defined functions) that add features to Teradata and support common SAS code. You should be able to use SAS and it automatically converted your work into SQL which is procesed by the Teradata system. The intention isn't to extract data out of Teradata and use it in a SAS system because that would greatly extend the time to process the data and get results. It would also lose a lot of advantages you have by using indexed tables and the processing power of the multi-parallel nature of Teradata.
My background is SPSS so I'm biased toward their tools, but any data analysis 'solution' should be designed to use the existing data storage infrastructure and not require data duplication.