Re: 2 million row database from hell! help a newbie in over his head.
Date: Fri, 24 Aug 2001 04:04:46 GMT
Message-ID: <ynkh7.12878$2u.96814_at_www.newsranger.com>
In article <X4jh7.185824$%a.7794082_at_news1.rdc1.sdca.home.com>, Carl says...
>
>Well, in my practice - we design the database for real world application.
>This means that it may be necessary to have some redundant data. This is
>not what I was talking about though. I was suggesting an aggregate table(s)
>to contain popular report data. We have used this quite a bit to transfer
>the load of performing popular aggregates to off peak hours.
>
>Just what is a "snapshot" index? I have never heard of this. What dbms are
>you speaking of?
My confusion, sorry. There is no such thing as "snapshot" index and there is little need too: I forgot about query rewrite. For example, a query
select sum(sal), deptno from emp group by deptno
could be speed up with a snapshot. Even though we could explicitly write fast query as
select sumsal, deptno from empsnapshot
substituting original query with something referencing auxiliary data structure in the application code is not nice. With query rewrite feature we still issue the original query, and the optimizer is able to find fast access path (via snapshot). In a sence snapshots with query rewrite feature are indistinguished of indexes. Received on Fri Aug 24 2001 - 06:04:46 CEST