Re: When should one rebuild an index?

On Dec 28, 7:08 pm, "Bob Jones" <em..._at_me.not> wrote:

snip

> Index structure and space usage are meaningless in determining index
> rebuild? So in your opinion what is meaningful?

Bob it is "getting to be" fairly well accepted in oracle dba land that for the most part the need to rebuild indexes on a regular scheduled ongoing basis is way oversold.

That being said, many shops have available cpu and time and schedules that allow scheduled rebuilds to not interfere with business priorities.

One can build test cases which either demonstrate that "little to no gains" to applications ( and important business activities ) are gained when doing certain rebuilds or "significant gains" are made.

Cary Millsap has a good definition that is relevant to meaningful if you want to read his book on Optimizing Oracle Performance. To badly paraphrase it the basic concept is when the net gain to the business of performing the performance tuning ( which an index rebuild is one potential type of a performance tuning method ( or at least an attempt at one )) is not much more than the cost of doing the performance tuning ... that's when to stop.

The guy you are posting back and forth with just loves to keep posting and will try to make strange points unrelated to your replies. Received on Sun Dec 28 2008 - 19:41:44 CST