Oracle FAQ | Your Portal to the Oracle Knowledge Grid |
![]() |
![]() |
Home -> Community -> Usenet -> c.d.o.server -> Re: how does CBO handle old statistics
It uses whatever it is told. The only time rule based is used is if it
is specified or if choose is specified and there are no stats on any
table in the query.
If the relative quantities of all your data is consistent over time, it may not make any difference. Some people advocate rarely updating the statistics for this reason.
Personally, I lean more towards gathering stats more often, along with having someone around to deal with any issues should they arise, such as bogus stats or the data passing an inflection point and the optimizer changing how things are accessed. See also plan stability, which would be preferred for an organization doing proper development, ie, checking the optimizer decisions for every piece of code before implementation.
It all depends on the type of system. A mixed OLTP and DSS system with RAD, extreme programming or a query generator, for example, wouldn't be able to check every piece of code before running, and would need someone around to put out fires. Even the most stable system can be punked by some newby using Access through ODBC or whatever.
jg
-- @home.com is bogus. I have eight projects spread out in front of me, and somehow I don't see most of them most of the time. http://money.cnn.com/2006/03/09/magazines/fortune/cubicle_howiwork_fortune/index.htm?cnn=yesReceived on Thu Mar 09 2006 - 18:27:28 CST
![]() |
![]() |