Archives of the TeradataForum

Message Posted: Mon, 09 Jul 2001 @ 16:08:36 GMT

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

<-- Anonymously Posted: Monday, July 09, 2001 11:47 -- >

On question 1 you ask about release of space in the mload_spool database. It sounds like you have a database set up specifically for capturing the work tables, error tables and restart logs. Teradata MultiLoad utility uses permanent space rather than spool space for work tables, error tables or restart log tables. In my work area, we calculate approximately 20-30% of the data space to accommodate such work files. There is a possibility that the volume of data being loaded has reached a threshhold that exceeds the amount of space you have set aside for these work tables.

The formula for preparing preliminary estmates for one target table, assuming no fallback protection, no journals and no NUSIs :

PERM * (using data size + 38) * (number of rows processed) * number of apply conditions satisfied) * (number of Teradata SQL statements within the applied DML).

If you are using fallback, multiply your estimated permanent space requirements by 2.

To estimate space required for changes to the NUSIs multiply (index lenght + 66) * (number of changes to the indexes) * ((1 for insert or delete) OR (2 for update)). (To account for fallback tables, multiply the NUSI Perm amount by 2. This amount would be impacted by conditional APPLY expressions and multiple NUSIs with different index lenths.

You would have to repeat the estimate for each target table and add space requirements for theerror tables -- two per target table. Multiload error tables have fallback protection and thier size varies with the using data size specification. BTW, adding more checkpoints is not a bad idea, but it will increase the number of inserts into the work tables (one for each).

There is more information about these space calculations in the Multiload Reference Manual.

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