DSXchange

Which Teradata Stage is more efficient when there is huge input data for extraction?Input volume:approximately 450 million recs.
which Teradata stage (Enterprise or API) helps extraction of data more efficiently?
Also after extraction from any of the above is done,i need to check if the extraction returns no rows then interface should abort.how can this be done?

Definitely Teradata (TD) API is not the choice if you have millions of records. TD EE (Enterprise) when used as source calls FASTEXPORT (TD Unload Utility). For explicit exports or big data pulls TD EE can work fine, or you can use the MLOAD stage (which does Fastexport if it is a source stage).

To check whether no rows were extracted before proceeding you need a modular design - one module (job) to perform the extraction into a staging area (Data Set), and a later module to process the data from that staging area. Perform your check before starting the second module.

Yeah,for the volume of input data to extract,modular design is suitable.but how do we go about checking if the extraction is empty.Taking a count of the table is not appropriate looking at the huge volume of data.Is there any other alternative?

Try selecting just the first row from the table.

TD API reads the records sequentially and is preferred when you have to read small number of records although it does not have any limits pertaining to TD utilities.


TD Ent stage is the only stage which can read and write in parallel and is preferred for large volume of data. it uses TeraRead and Terawrite or in other words fast export and load protocols and thus has the same limitations which the TD utilities have.

Teradata Certfied Master V2R5

Interesting... I think TD API have the option to run in parallel mode also. One can write/insert records using TD API in parallel mode and it gives much better performance than using the same in sequential mode.

hamzaqk wrote:TD API reads the records sequentially and is preferred when you have to read small number of records although it does not have any limits pertaining to TD utilities.

TD EE is not the 'only' parallel stage I think. TD Mload is also widely used.

hamzaqk wrote: TD Ent stage is the only stage which can read and write in parallel and is preferred for large volume of data. it uses TeraRead and Terawrite or in other words fast export and load protocols and thus has the same limitations which the TD utilities have.