Single RBS for DS system

From: Lee Parsons <lparsons_at_eskimo.com>
Date: 1995/04/20
Message-ID: <D7CtoD.JHv_at_eskimo.com>#1/1


I'm implementing a Decision Support system next week I had a quick question about rollbacks segs.

The system in question is a 5GB DB that loads data from legacy sources and provides single point access for managment reporting. The DB is refreshed on the following schedule. 50% monthly 25% weekly and 25% daily. We expect no more than 50 concurrent users. Nightly loads are done by a single batch process and the only updated that are allowed are done by the load process immediatly after the bulk loads. We use sqlldr in direct mode to populate the database.

After looking over our environment and thinking the problem through I have decide that the best roll back configuration here is one large RBS. Since individual's wont have access to change data and only one process at a time will be allowed to do updates it seems a bit silly to have a Big RBS and a bunch of small RBS and have to worry about which one the load process is using. Rather, I like the idea of having one 50M RBS.

Since this flies in the face of popular (documented) wisdom I have to ask:

Reasonable?
Any DS impementors out there that went this route and regretted it?

Remember we have 50 query users and the only updates are done by a single process with no other users on.

-- 
Regards, 

Lee E. Parsons                  		
Systems Oracle DBA	 			lparsons_at_world.std.com
Received on Thu Apr 20 1995 - 00:00:00 CEST

Original text of this message