Archives of the TeradataForum
Message Posted: Wed, 23 Mar 2005 @ 12:18:54 GMT
| Subj: || || Re: Purging Data from tables |
| From: || || Dieter Noeth |
Naveen Ram PrasannaKambhoji wrote:
| ||1. Tables should be cleaned up once in a month and Volume is too huge around 50 Million records I need to delete.|| |
| ||2. Window for the maintainance is 2 - Half peek Hours.|| |
| ||4.Currently data already exist in the table and the applications/reports make extensive use of this table so we defined PI on
different set of Columns.|| |
| ||5. I came to know that we cant use a Multi Load delete because dateid is not a PI and MultiLoad delete will work only with
WHERE condition on PI Columns.|| |
You can't use a BEGIN [IMPORT] MLOAD, but a BEGIN DELETE MLOAD: Faster than a SQL delete because there's no Transient Journal and no rollback
in case of failure (just restart the job).
Is your DateID column a date or an integer? Then you should think about a Partitioned PI.
But do some test before: This may boost your deletes ans some queries, but other queries may slow down, it depends...