[SIPForum-techwg] SIPConnect 1.1: REGISTER support MUST at SIP-PBX

[Johnston, Alan B (Alan)]

Theo,

Thanks for your comments.  See mine below marked AJ>

Thanks,
Alan

-----Original Message-----
From: Theo Zourzouvillys [mailto:theo at crazygreek.co.uk] 
Sent: Friday, January 30, 2009 3:24 PM
To: Johnston, Alan B (Alan)
Cc: techwg at sipforum.org
Subject: Re: [SIPForum-techwg] SIPConnect 1.1: REGISTER support MUST at
SIP-PBX

On Fri, Jan 30, 2009 at 7:34 PM, Johnston, Alan B (Alan)
<abjohnston at avaya.com> wrote:

> implies an incorrect master/slave relationship

In what way does using REGISTER create a master/slave relationship
that other rendezvous methods don't (p2p excluded, i assume :))?

AJ>  I don't mean P2P but I do mean peering.  The real model between a
service provider and a PBX is a peering relationship, and peers don't
register with each other.    

> There
> are other solutions such as dynamic DNS for situations where this can
> not be achieved.

I fail see the differences between REGISTER and dynamic DNS - they
both archive identical things - updating a database?

anyhow, how would this be as a solution to what you believe is the
pathological NAT & dynamic IP address cases in a standards and optimal
way.

 a SIP-PBX MUST support TLS, and MUST be configurable to always keeps
an active mutually TLS authenticated connection open, which it will
re-open if closed.

... nice and easy for the SIP-PBX to implement, and then the SSP can
do all the grunt work to make it work in real world deployments :-)

AJ>  I am not suggesting dynamic DNS as a real solution to this problem
- I'm just pointing out that the NAT problem is not unique to Internet
Communications, and there may be other solutions at other layers of the
stack.  I agree in the presence of NAT it would be useful to maintain a
connection between the Service Provider and IP-PBX.  I'm not sure how to
specify this exactly.


 ~ Theo