I just read https://activedirectorypro.com/group-policy-best-practices/
and thought it contained a pretty good idea.
Rather than using Deny, it encourages that "GPO1" be applied at the OU root level and then creates a sub-OU for which a contradicting "GPO2" is applied (directly).
But, after thinking about it, one realizes that things can only be a set of unique objects in each OU.
So then, what if one has another
treatment to apply via "GPO3" to a set of objects that overlaps the membership of "GPO2" above?
Then the method falls apart.
I've often wondered about this and am looking for some logical structure that would allow:
GPO1 applied to a,b,c,d,e,f,g
GPO2 applied to f,g
GPO3 applies to e,f,g,h
There is an intersection between all 3 GPOs: f,g
There is an intersection between GPO1 and GPO3: e,f,g
GPO3 has a member (h) that is not in the other two GPOs.
In my cases, so far, I find that there would be a GPO to apply to all Users or all Computers with some exceptions in each case.
So, creating OUs for the exceptions as suggested, may likely not work.
There only needs to be yet one more GPO with a different set of excepted objects and with overlap, but not 100% overlap, of those objects.
The referenced note above suggests one way when one OU is part of a root (or higher level) OU. But that seems to barely scratch the surface.
What is the best way to tackle this rather obvious situation/objective in designing GPOs, OUs, etc?