Re: 2 million row database from hell! help a newbie in over his head.

In article <_NUg7.298$Iw2.19331_at_petpeeve.ziplink.net>, David Cressey says...
>
>> more relational) solution would be using index. I'm not sure if built-in
 indexes
>> for "group by" are readily available yet off the shelf, but with extending
>> indexing feature you can certainly program one.
>
>You've lost me, here. I guess I've never worked on an RDBMS where an
>appropriate index would NOT
>be used by the optimizer to implement "group by". As far as I can tell,

David, I really meant aggregates. Suppose we have a query

select avg(SAL) from EMP

Is there built-in index that would make this query fast? With group by? I was assuming that the whole data warehousing mumble has been around because databases are so index-challenged.

>Oracle uses an index on the group by column with both RBO and CBO.

I was not aware of that. I'll experiment with plans to learn...

>And I'm sure that the index is used with Oracle Rdb (not to be confused with
>Oracle RDBMS).
>
>What environment are you discussing where you have to "program your own"
>index?

8.1.7 Received on Wed Aug 22 2001 - 23:41:50 CEST