RE: BEEP Transport

["Hollenbeck, Scott" <shollenbeck@verisign.com>]

> -----Original Message-----
> From: Eric Brunner [mailto:brunner@nic-naa.net]
> Sent: Tuesday, September 04, 2001 4:28 PM
> To: Hollenbeck, Scott
> Cc: 'ietf-provreg@cafax.se'
> Subject: Re: BEEP Transport 
> 
> 
> 
> > BEEP does not have an explicit dependency on TCP, though this draft
> > describes TCP connectivity.  Would it be possible to clarify either a
> > specific requirement for TCP transport (expanding on the reference to
> > RFC3081), or note that protocols other than TCP are possible and that
this
> > is just one possible way of moving data around?
> 
> There is a reference to 3080, as well as to 3081. I don't see the problem,
> but do suggest language.

I know this is a very small nit, but I think it important to note that the
requirement for TCP/3081 comes from this document and not from BEEP itself.
A statement to make that clear may answer future questions like "why didn't
they use SCTP?".

I can't provide text to explain why TCP was chosen, though I'll admit that
it's an obvious choice given that it does the job and it's documented,
whereas other transport options are still immature or undefined.  I can
suggest text to let folks know where the requirement is coming from
(reference is paragraph two of section 1, "Introduction").  Please consider
changing this:

"This memo documents how a EPP session is mapped onto a single [BEEP]
session.  [RFC3081] documents how a BEEP session is mapped onto a single TCP
[RFC793] connection."

to this:

"This memo documents how an EPP session is mapped onto a single [BEEP]
session.  [RFC3081] documents how a BEEP session is mapped onto a single TCP
[RFC793] connection.  This profile requires a reliable, connection-oriented
transport that preserves the order of data exchanged between end points.
Though TCP is not specifically required by BEEP, TCP provides the service
attributes required by this profile."

<Scott/>