Re: multiple schema or multiple instance

On Thu, 30 Sep 2004 23:18:52 +0200, Bruno Jargot wrote:

> But the key point is the restore of the database. If a schema require a
> incomplete recovery,

Does that mean that you predict schema which would require losing some data? Why would you enter such data in the 1st place? If the data was entered by mistake, then you should have application standards so that you can resolve things. Every table that will undergo updates, for instance, should have at least 4 columns: date_created,date_modified,user who did the last modification and type of the modification. Sometimes, you also want to add version number and the current flag. In other words, creating a proper data model will save you resources. Each instance needs its own SGA, background processes, semaphores, service directories and all other overhead that comes with an instance. As for the instance recovery, in case of a damaged datafile, it can be done without impacting other users. Having may databases wastes machine resources and complicates database administration.

-- 
Artificial Intelligence is no match for natural stupidity.