Group IDs are used to make directory servers preferentially choose replication servers of the same group ID. As such, they can be seen as a way to increase performance by reducing networking latencies. As a consequence a sensible way to use them is to map group IDs should map to physical locations like data centers.
In 6.5 separate group IDs are assigned for each replication domain. This does not make sense, because a server is in a single physical location, independent of the replication domain.
This task can be closed once:
- servers are only tagged with a single group ID (the global group ID)
- 6.5 servers can be upgraded gracefully