 |
 |
Archives of the TeradataForum
Message Posted: Mon, 25 Jul 2005 @ 15:09:48 GMT
| Subj: | | Nested summary tables |
| |
| From: | | Aananth.S |
HI All,
This question is with regard to the Join indexes that TD touts as the solution for summary tables. We have similar features in oracle
(materialized views -- alebit lots of restrictions on refresh and performance). But the one good thing about Oracle MVs is that we can have a
nested structure. How we generally go about this is that we take the fact table and agreagate it using a few dimensions and build a MV (Aggregate
MV or Join MV or a combination) and then using this MV, we go on to summarize fruther using other dimensions. Now, in TD is there a possibility to
have such a structure. In our sales Datamart, we have close to 7 dimensions for a fact table. Oracle does not scale if we were to join everything
in one MV -- close to 2 billion rows as a result of such a join with just a years data!! My question is how do we implement summarizations in TD
in such a scenario. It scares me to think of aggregating all 7 dimensions into a single JI (agg) !! what would you experts generally recommend to
do?? Yours suggestions are welcome!
Thanks
Aananth
| |
