Archives of the TeradataForum
Message Posted: Fri, 26 Mar 2004 @ 15:58:05 GMT
Although what you say about reading the existing data and sorting it sounds easy, it would require that the system have enough PERM disk space to hold all of the original table rows, these existing rows again in a work table and the new rows being loaded. Besides the speed factor, this could easily exceed 3 times the original table size.
As it is, once the new data is loaded into work blocks and sorted, all the AMPs need do is write the work blocks into table blocks as fast as they find available sectors and write them.
Just another perspective and regards,
|Copyright 2016 - All Rights Reserved|
|Last Modified: 27 Dec 2016|