Hi Dieter,
This is my first post and I am have been following you for a while before creating an account. Thanks for your valuable suggestions.
Request you to clarify me on the below.
1. Does the join query dependent on number of AMP's because when the same query ran on less number of AMP system, the small table has been redistributed, but when it is ran on large AMP system, the larger table has been redistributed.
If not, then what forced teradata to distribute different tables in both scenarios.
2. As long as tableb has the UPI on the join column, it won't be redistributed!
As, Ayush mentioned that small tableb with UPI has been distributed across all AMP's eventhough it has an UPI.
How does it happen? (Both are contrasting according to me atleast!)
I hope that redistribution and duplication of small table is entirely different.
3. You have suggested to use "col*1" or "coalesce(col1,col1)" to help optimizer to ignore the index.
Now, the 2 tables has to be distributed as it can't use the index of both the tables right? which will even degrade the performance.
Please enlighten me on the above. Excuse me for any silly question.
Thanks,
Vikram.
↧