Re: Last Call: <draft-ietf-tsvwg-iana-ports-09.txt> (Internet Assigned Numbers Authority (IANA) Procedures for the Management of the Service Name and Transport Protocol Port Number Registry) to BCP

[Cullen Jennings <[e-mail]>]

So to summarize what you are saying, ports are allocated based on an arbitrary view of the expert review. When this person will say yes or no too can't be described and will change over time. 

If that's how it works, there is not even any grounds for appeal of any given decision. You can't even use precedence as an argument. My view was the IESG has been trying to move to having much more concrete advice for registries that required expert review. If you agree that is about the right summary, then I'm pretty sure I can find plenty of other people that would agree with me that this is not OK. I'm not a fan of the WG could not get consensus on if we should allow A or not so we are just going to let the expert review do whatever they want. If the IETF could not come to consensus on if X  or Y were reasons to deny a registration, then I don't think the expert review should be able to deny a registration due to X or Y. 



On Jan 31, 2011, at 8:06 , Magnus Westerlund wrote:

> Cullen Jennings skrev 2011-01-30 05:56:
> >
> > I read the draft to say that there would only be one port allocated - I took strive to mean that Joe would deny my port requests for two ports. If the intention is actually for the draft to say that it strives for one port but allows assignment of two where the that is what the protocol design desire, then I have no problem. Perhaps we just need to clarify what "strive" means. This definition of "strive" leads into exactly my other complain that this draft provides no guidance on what the expert will or will not approve.
> >
> > We probably need to adjust text like
> >
> > o  IANA strives to encourage the deployment of secure protocols, and
> >      so strives to avoid separate assignments for non-secure variants
> >
> > and
> >
> >  The use of separate
> >   service name or port number assignments for secure and insecure
> >   variants of the same service is to be avoided in order to discourage
> >   the deployment of insecure services.
> >
> > and
> >
> >  Services are expected to include support for security, either as
> >   default or dynamically negotiated in-band.
> >
> >
> > In band negotiation of security is applicable for some cases, but it adds latency, bandwidth, and complicated multiplexing in non session based transports. I think this is a bad idea in many cases. I also view separation even for stream based protocols as something that helps management and debugging as well as policy.
> >
> 
> Well, the high level goal is to preserve a limited resource. We can't do
> that without making the preference clear. But I think you have forgotten
> to consider those statements trying to make clear that this is up to
> review.
> 
> The review criterias can be expected to change overtime. They are also
> hard to codify. Especially for this case, how do we measure
> architectural uncleanness, implementation issues, and performance impact
> to make a rule that judges if one or more port is allowed?