 |
 |
Archives of the TeradataForum
Message Posted: Wed, 03 Dec 2003 @ 11:03:51 GMT
| Subj: | | Re: Fastload Acquisition Phase |
| |
| From: | | Anomy Anom |
<-- Anonymously Posted: Wednesday, December 03, 2003 05:51 -->
Hi,
Yes.. I understand that it does transfer the data in blocks.. But when it transfers to the AMPs, how would it know, which block to
go to which AMP..? Some hashing principle should be there for distributing the block of rows into the different AMPs, I guess..
Similar thing happens in Multiload phases too.. It first puts the rows into the work tables which are situated in different AMPs..
How would teradata know which block/row to send to which AMP before the receiving AMP hashes it on primary index and redistributes it
to the correct target AMPs. ? Again some distribution mechanism should be there to split the data across the work tables, before they
get applied to the actual target tables.. Is it not..? if it is, then what is it..?
Cheers
| |
