Re: [Sip] Question on loop detection

["Elwell, John" <john.elwell at siemens.com>]

Theo,

Thanks. That is indeed one solution, but I am not sure all B2BUAs
deployable in such situations behave in this way, so I would welcome
other opinions.

John

> -----Original Message-----
> From: theo at crazygreek.co.uk [mailto:theo at crazygreek.co.uk] On 
> Behalf Of Theo Zourzouvillys
> Sent: 04 February 2009 17:00
> To: Elwell, John
> Cc: sip at ietf.org; Armenio, Joao
> Subject: Re: [Sip] Question on loop detection
> 
> On Wed, Feb 4, 2009 at 4:31 PM, Elwell, John 
> <john.elwell at siemens.com> wrote:
> 
> > Questions: Has this problem has been seen in practice? If 
> so, what steps
> > have been taken to overcome it? If not, have I 
> misinterpreted RFC 3261?
> 
> The behaviour as you described it is correct when the SBC device is a
> B2BUA with a single UA core for all sessions flowing through it, which
> it isn't - or at least probably isn't what you want.
> 
> if instead you considered an SBC to have multiple logical UA cores,
> one for each target that it's currently being an SBC for, then that
> behaviour would be incorrect (as matching would be done by comparing
> all other ongoing transactions within the same UAS core), and both
> calls would flow through correctly.
> 
> this is how we have had our SBC behaving for the last 3 years without
> problem [1].
> 
>  ~ Theo
> 
> 1 - standard SBCs don't exist and it's all in our heads 
> disclaimer applies :)
> 
> -- 
> Theo Zourzouvillys
> Chief Technical Officer
> VoIP.co.uk - Commerce House, Telford Road, Bicester, OX26 4LD
> Tel: +44 1908 764 196
>