Re: Proving or Debunking the need for rebuilding

On Wed, 15 Nov 2006 07:42:37 +0000, Richard Foote wrote:

> Now this could be attributed to an index rebuild but then again it might
> not. For example, I assume you re-analyze the index after the rebuild ? What
> are the differences in index statistics before/after the rebuild ? When was
> the index last analyzed prior to the rebuild ? There was one classic thread
> here a while ago where Mike Ault swore an index rebuild changed the
> clustering factor of an index only to realize (eventually and somewhat
> painfully) that the rebuild made no difference to the CF, it was simply the
> statistics being no longer fully up to date. Is it possible that by simply
> re-analyzing the index/table, performance improves without the need for the
> rebuild ? I've seen this on various occasions ...

There is also a classic presentation about indexes on OAUG now moved to Miracleas.dk:
http://www.miracleas.dk/images/upload/Docs/Richard%20Foote.pdf I find the myth debunking section hilarious. That presentation explained CF to me long before Jonathan's book.

-- 
http://www.mladen-gogala.com