Discussion Locked

How do you recover distributed databases

0 helpful
3 replies
Closed to new repliesThis discussion is closed. It stays readable, but it takes no new replies.

I've been asked to define our recovery strategy for a Global application. The application consists of multiple Oracle databases.

The technical people have advised me that the technology alone cannot guarantee the databases be recovered to a synchronous point in time.

This leaves the application team to write programs and utilities to reconcile the multiple databases etc.

Could you please tell me how your recovery strategy works for multiple Oracle databases.

Thanks

3 replies

by tony derosa

How do you keep the global DB up to date without replicating data from the regional sites? Somewhere, somehow the regional transactions need to be reproduced on the global, or at very least the current tables have to be periodically pushed up to the global or pulled from the regionals and merged, right?

Also, do the regional sites get the common data directly from the global or do they keep a local copy? A local copy would need update as well from the global.
Can an application modify regional data in the global DB? This means updating the regional site.

You can create schemas and views to support peer to peer replication, or at least implement a master/slave replication where one way update is sufficient.

Tony

by jenx

Thanks for your answer Tony. You're right, I need to expand my question a bit so that people can visualise the situation clearer.

The application is made up of multiple databases - one is a single global database and the other databases(min. of 2), will occur for each region that cuts over to use the application. (We're not going big-bang.)

Data that is stored in the global database is a combination of common data and different data, from that stored in the regional databases. (Therefore, we're not able to use Oracle replication.)

The recovery strategy needs to solve how the application can recover itself if one of the databases, for example, becomes corrupt.

We need to have a solution that makes each of the databases consistent with each other i.e. up to a specific transaction.

The technical team are looking at reducing the risk by adding mirroring on the global layer, in the aim to reduce data loss. However, we still have to build an application recovery solution.
The solution that I'm looking at is building reconciliation reports and programs that will recreate the data from the Global database.(If data isn't present in the global database that is required in the regional database, then I'm looking at defaulting this to dummy values.)

Do you have any thoughts on my proposal?

Thanks
Jenny

This discussion is closed to new replies.