Archives of the TeradataForum

Message Posted: Fri, 21 Mar 2008 @ 17:00:44 GMT

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

Are you getting this error out of a 3rd party ETL or BI tool? If you are, and if you are applying significant amounts of data to the database using row-at-a-time INSERTs, UPDATEs or DELETEs by a 3rd party tool in ANSI mode, you may want to consider using a different mechanism to load data.

For example, the DataStage 7.5.x SE Teradata API Stage performs row-at-a-time operations against Teradata in ANSI mode. Using this stage, we can avoid getting this error if we set the commit interval to every 1000-2000 rows instead of waiting until the end of a load to commit. However, commiting during the load makes process restartability difficult (unless you had a tool like [i.e. FastLoad & MultiLoad?] where the entire tool is aware of when checkpoints are taken / commits are applied ...)

We currently ask our development teams not to use row-at-a-time load operations if they are applying more than 2000 rows. We've found some other loading options have allowed us to improve load time, increase the granularity of our DBQL logging, and improve overall system performance without impacting clean restartability.

- Chris

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