 |
 |
Archives of the TeradataForum
Message Posted: Thu, 21 May 2009 @ 15:10:23 GMT
| Subj: | | Re: Column Order in Derived Table |
| |
| From: | | Dieter Noeth |
Srikanth Goli wrote:
> SEL ....
> FROM A
> Join (Sel
> From B
> Where ..
> )C
> ON A.Col1 = B.Col1
| | Does it use the first column in the derived table B as the PI OR does the optimizer understand that regardless of location of B.Col1, that
needs to be driver for distribution / hashing? I understand that with standalone "Create Tables As", this is a factor and that first column either
be the PI column, or PI be explicitly specified. But with a derived table, where optimizer has access to statistics and query, does it still need
the user specify the order? | |
No, the order of columns shouldn't matter.
In TD12's Explain you'll see on which column(s) a spool is redistributed. Before TD12 you need to submit a "DIAGNOSTIC VERBOSEEXPLAIN ON FOR
SESSION" to get that detail in explain.
Dieter
| |
