backup strategy

From: Audun Jensen <audun_j_at_yahoo.no>
Date: 23 Feb 2001 15:33:45 GMT
Message-ID: <Xns9051A87421E72audun.jyahoo.no@130.133.1.4>

Hi,
we have an application currently running on MS SQL Server, and now we want to support Oracle as well.
This application create SQL server databases "on the fly", on oracle it will create schemas "on the fly". These schemas will consist of a customer's project. An instance can contain "thousands" of schemas.

I'm wondering what approach to take regarding backup/recovery. I read that physical backups contains physical units like datafiles, tablespaces etc. There is also a logical type that uses exp/imp on user leve. Which one is most common? Why?

Should the schemas created on the fly put it's objects on a separate tablespace? In that case the physical backup strategy could be used, and a customer's project could easily recovered. If so, the instance will contain thousands of tablespaces (among with thousands of users with approx. 50 tables each). How will Oracle handle this?

Anyone who could give some advices?

AJ Received on Fri Feb 23 2001 - 09:33:45 CST