Archives of the TeradataForum

Message Posted: Thu, 05 Dec 2002 @ 16:49:20 GMT

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

We do a lot of transformations on Teradata. Including examples like:

Converting 30,000 lines of C code to apply inventory status into 15 step MLOAD, SQL process which runs in minutes.

Householding 180 Million individuals into 90 Million households using 16 different business rules with between 2 and 5 data attributes in per business rule using thousands of SQL requests and applying these to the warehouse tables in less than 2 hours.

Teradata; however, is not your average RDBMS. It takes some knowledge of how to do these tasks and if you have complex requests or very large volumes relative to the machine, a little more knowledge how to do them optimally. Thinking in sets is very important for this process and people used to row by row processing sometimes struggle with this change in approach.

For example, You mentioned using Stored procedures. If this assumes using a cursor, I would guide against this serialized approach unless the volume is very small probably 0,000's of records or less. If you intend to use standard SQL requests in your Stored Procedure without cursors then you can take full advantage of Teradata's Parallel processing to greatly speed up transformations.

As far using this approach with ETL tools, the tool vendors really need to take a different approach with Teradata and only a few have. Mr. Fortenberry mentions one of these.

Many try to treat Teradata like the rest of the RDBMS. This will work pretty good for normal volumes if you streamline certain standard approaches like the ones mentioned by Mr. Judge.

However, to optimize large, complex, and high volume transformations, you'll need a tool that understands Teradata or you may need to write the transformations yourself.

Thanks,

Eric

EJK
Eric J. Kohut
Senior Solutions Consultant - Teradata Solutions Group - Retail
Certified Teradata Master
NCR Corp.

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