|
Archives of the TeradataForumMessage Posted: Wed, 06 Jul 2005 @ 14:01:45 GMT
Thank all the guys who responded my emails. It was a very good discussion. I will try to use Teradata utilities to load data as many recommended. However, rollback took so long time vs. INSERT is not acceptable. If you have a transaction involving many tables and data integrity is a major concern, SQL is the best choice without doubt. I do not think Teradata should keep itself as it is now (How much better V2R6 is?). I totally understand Teradata is specially designed for TERABYTE database. In my first email , I said I had been ready for brain washing. I cannot divorce Teradata. I have to know its behaviors, good or bad, to work with it. I hope people here can provide new Teradata users like me more valuable experiences with Teradata. I know we have many very good experts who are willing to help. When Prasanna responded to one of my emails, he said, "Sometimes this can lead to a deadlock with queries fighting for resources/spool space etc". I have concerns too. Teradata is using 32 bit CPUs. What about 64 bit CPU? It has been in the market for a while. The memory addressing space is almost unlimited for 64 bit CPUs. And the price of memory is going down steadily. If we can afford to put 1 Terabyte memory in a box, does the AMPs, heart of Teradata, are something you still need? Data, once read into the memory, can stay in memory, disk I/O may not be an issue. And data can be shared by many virtual/real processors, not a unit like one AMP. What about bynet at that time, still needed? Maybe for master/slave pairs. I am talking about its future. I was told that NCR is working on 64 bit systems. Thanks, Charles
| ||||||||||||||||||||||||||||||||||||||||||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||
Copyright 2016 - All Rights Reserved | ||||||||||||||||||||||||||||||||||||||||||||||||
Last Modified: 15 Jun 2023 | ||||||||||||||||||||||||||||||||||||||||||||||||