This is a bit of a stab in the dark, but I was wondering if anyone here has been involved in creating a database replication algorithm. We need to come up with something in the next few months and we'd really like to avoid re-inventing what could be a quite tricky wheel.
To give some background, we need a geographically distributed database behind our application. Under guaranteed connectivity, we can arrange to have all writes to be made to a master database which then hot-copies to a collection of slaves. Unfortunately, the connectivity cannot be guaranteed, but local databases are still required. So some sort of replication will be required. (We already know we will need the application backend to manage it as it will have the knowledge to sort out conflicts that the database won't have.)
I've started Googling, but the IWETHEY Group Mind would be useful to plumb, too.
Wade.