An issue which may be of concern to any DBA working with single and dual-target database replication is the ability to move between dual targets and single targets, and back.
Dropping a target in a dual-target world may mean that the source database may be in jeopardy until the database is shut down. The AIs may continue to switch until all are full, and then, the source database will require some form of emergency maintenance.
A failure of the source database in a dual-target environment may require a DBA to revert quickly to a target system. This new source system may have to be converted into a single-target source. In this case, the remaining target database would be unusable and would need to be restored to establish a single target environment.
In environments where a source and target database are required 7x24, this can be an issue. What this means is, in a failover situation, the target is always lost until restored. A DBA cannot re-point synchronized databases to new targets at-will.
If you are a DBA in a multi-GIG dual-target database environment with a dead source host, this can mean significant down time while a new source (old target) database is backed up, and restored to the remaining target. There is NO quick failover.
Host maintenance in this situation usually means some sort of outtage or jeopardy, just as if the host had failed.
What awareness exists in the Progress community of this situation, and does anyone feel that this behavior merits some sort of redesign?
When all three databases are in sync, it would be preferrable to simply at-will repoint any database to any other single database or pair of databases. Would it not?