Enterprise Design Back to Front

Hi all,

We are preparing to go live with a major redesign of our suite of applications and I am seeing some "bumps" on the horizon. As you know the SDLC process is more something that is aspired to rather than something that is practiced. The project is over two years behind and many roadblocks and reversals were experienced. Our suite involves many different technologies. We have tried to reduce the complexity, modernize and update. We are centralizing operations of about 20 different business units with their own idiosyncrasies and ideas into one massive operation. Of course all our business units are against the idea and we experience the level of cooperation that is commiserate with that attitude.

The suite contains fourteen or more databases will be incorporated into one transactional database and reporting activities shuttled off to a data warehouse. We are starting off with ten terabytes of data. Some of the business operations are pretty small and some are, at least to me, very large. We will begin with a clean system and import each of our business units into it one at a time. The database is virtual private database (VPD) enabled and over time it is going to get quite complicated as each activity joins, processing and interchange increase. We don't have a test system or test data so it has not been properly load tested.

The new design has a single Solaris database server. I am being asked to come up with an operations, maintenance, and backup schedule for this system. The customer is not overly familiar with the concept of High Availability. Every time I say "High Availability" they say "Disaster Recovery." They apparently have funding to build a disaster recovery site somewhere in the Midwest. They think this will allow us to do rolling upgrades. I'm not terribly experienced at this type of operation but at least I recognize the flow chart. I don't think this will work.

Is it possible to use a disaster recovery site 700 miles away from your primary site which is four hundred miles away from our office to manage maintenance? I'm thinking we need RAC and a physical standby in the same location to do this. We are never going to have more than a weekend to do maintenance. They have been quite tolerant of downtime in their various centers but I think it will be quite different to experience an outage that affects all of their 60,000 employees. I'm thinking "15 minutes" is about what we should be striving for in terms of time to restoration. It seems difficult to nail down SLA's and all our documents are labeled "Draft."

So, I'm thinking it's impossible to use the disaster recovery site, which won't be available until months after go live, to do this.

--
http://www.freelists.org/webpage/oracle-l