 |
 |
Archives of the TeradataForum
Message Posted: Fri, 25 Jun 2004 @ 09:19:16 GMT
| Subj: | | Re: GLOBAL Vs LOCAL Aggregation |
| |
| From: | | Victor Sokovin |
Prasanna,
| | We have a situation where we perform a join on small table(882 ROWS) with a very huge table(130 Million Rows) for doing an aggregation i.e
we do a roll up from daily to weekly & monthly values. When I went through the explain only the SUM step is taking > 1 Hour. Currently we using
v2r5 with 20 AMPS.Is it possible to achieve performance on summation step. I have all the STATISTICS defined on the table. Suggestions and Inputs
are most welcome | |
| | Also I would like to know about this GLOBAL & LOCAL Aggregation technique, all-AMP and Cross-AMP Aggregation technique if they exist
please do let me know!! | |
I believe you had posted the same question a few days ago. It would appear that there were a few suggestions made. Because your query is so
simple and basic there is not much, in my opinion at least, one can do with it as such on a 20-AMP system. The run time is pretty much dictated by
the data volumes and the hardware.
Basically, I see two options for improving performance in such cases:
Option 1: add more AMPs/nodes;
Option 2: rewrite the query using suggestions and ideas in the previous thread.
Regards,
Victor
| |
