During Working Group call we agreed that we need to clarify how non-transparent redundancy is expected to work.
Either:
1) The ApplicationUri is used to describe the Server set with each member of the set using the same Uri
2) or, there needs to be a new GDS way to associate servers in the Server Set.

Servers in a non-transparent redundancy set have unique ApplicationUris for each server in the set. Otherwise the reduancy information in the server and the discovery would not work.

Therefore each server in the redundant set needs to be registered with the GDS.
What we miss is a linking of redundant servers in the GDS but we did not publish an application information model in Part 12 (this was part of an early draft)

I think we have made a mistake sometime over the years by assuming the ApplicationUri and the ServerUri are the same. While this is true for a non-redundant server I think our original intent was to use the ApplicationUri to represent the Server Set of non-transparent redundant servers and the ServerUri to represent the individual members of the Set.
If this wasn't the case then what is the reason for having both ApplicationUri and ServerUri as separate concepts if they are always the same?

Consider using the Alias concept for Server Groups.

Explicitly defined behavior for that non-transparent redundant Servers in 6.3.6.

Need to clone to Part 4 to require non-transparent redundant servers shall register the Server Capability "NTRS".

Agreed to changes in Draft 24 edited in Virtual F2F.