Purpose of Separate Configuration Directory

Question

I know that Sun best practices recommend that the configuration directory instance be separated (placed on a different physical server) in a directory environment.
Other than the ability to manage all directories in a single administration domain, what does this accomplish?
Is there anything in a replicated multi-master environment that would require such a setup, with all master and consumer instances pointing to the same configuration directory?
Is it possible to set up a replicated multi-master environment with configuration directories on each individual server?
When do "user" directories need to access the configuration directory instance, other than during initial setup? If the configuration directory is on the WAN relative to a user directory, could inability to access the config directory prevent the user directory from starting/stopping/accepting requests/otherwise functioning?
Thanks in advance.

Answer 1

I don't disagree, however I think you can go too far. If everything is in one o=netscapeRoot, it's like putting all your eggs in one basket. If you loose that directory, you loose console access on all SunOne instances. And while having a multi-master config directory aleviates the problem somewhat, usually loosing the config directory does not occurr because of a hardware problem - rather user error or an upgrade gone bad - and the error usually replicates just fine. Backups are key before any operations that changes or could potentially chage the config instance. Personally I like to have one config instance per <Server Root>. It does not take much in resources.