Archives of the TeradataForum
Message Posted: Tue, 27 Mar 2007 @ 12:38:20 GMT
| Subj: | | Re: NULLs in the file to Mloaded |
| |
| From: | | D silva, Joseph Vinish |
| | We are loading a table having a NUPI. The load ready file is having 90% nulls in this Primary Index column. | |
| | EXEC MLOAD statement in Mload is taking huge time. Number of records in the file are 3 million. | |
| | Does the NULL in the file is the culprit? Please advice. | |
It's definitely the culprit. If your PI column is having 90% nulls, then you are also having a very bad skew (Atleast 90% of all records would
go to the same AMP) which means premature database full condition, trouble with spool space etc.
This is what one might call a bad PI choice, is there any reason why this column is chosen as a PI ?
Also for set tables with NUPI, having too much of duplicate values (in this case null) would result in TD doing lots of duplicate record
checking...
So in your case the trouble is double headed...
Joseph D'silva
| 
