[Nml-wg] NML terminology: Port vs Interface

[Jason Zurawski]

Hi Jeroen/All;

Comments inline:

On 9/27/11 5:03 PM, thus spake Jeroen van der Ham:
> Hello,
>
> My final 2 cents on the debate:
>
> The issue was raised in two separate projects which are looking to
> adopt NML for infrastructure description. These projects aim to have
> users as active writers of topology descriptions, not just passive
> consumers of measured data. The confusion is that these users are
> more accustomed to the use of "port" in the context of a UDP or TCP
> port, not as a network attachment point.

As I have noted in prior exchanges in email and in person OGF - I don't 
see this as a compelling reason for us to run out and change our 
established terminology.  What happens when another group approaches you 
at an event in another 2 years to call the concept 'port', will we have 
yet another discussion on this mailing list to change (again)?

I find this entire line of conversation rather unnecessary and 
counterproductive to this working group.  Its a *name*, and if several 
projects that initially called the concept an 'interface' can get used 
to 'port' re: Martin's earlier history laden email, I don't see why 
these new groups cannot do the same.  It is ok to tell people 'no' 
occasionally, especially when there is a destructive nature to the 
overall choice that will force work on many established partners.

> The concept of a UDP or TCP port would actually probably translate to
> a Label and not a Port/Interface if we would use it in a network
> description.
>
> Looking at GENI, and especially ProtoGENI as the leading candidate,
> they seem to go for Interface as well.

This is false.  ProtoGENI (something that Martin, the UDel/IU students, 
as we as Internet2 are heavily involved with) based their early work on 
NM/NML schematic concepts, mostly from old examples that featured port. 
  It is true its not a one to one mapping, but they are using links and 
ports just like we are.  Other GENI efforts (orca/ben, planetlab) may 
use interface, I dont know those as well, but these do not represent 
*all* of GENI.

Thanks;

-jason

> That being said, the world will not end if we choose Port. Those
> projects will probably use Interface anyway, but just semantically
> match it to nml:Port if necessary.
>
> Jeroen.