[Standards-JIG] Jingle vs. Zoep

Brian Raymond brian.raymond at je.jfcom.mil
Thu Feb 16 20:31:23 UTC 2006


This is quite a long thread so I¹m not sure where to jump in but I will
expand on my earlier comment regarding how some protocols do it in the
signaling path and why they do. I come mainly from an H.323 background with
H.320, H.321, SIP thrown in for flavor. I have written both server side and
client side code in these environments and have seen valid uses for both
sending tones in the signaling path, as well as in the media path. In H.323
I have seen tones, or other data as one post indicates (in my case simple
chat) sent in H.245 user input indication messages, or even Q.931 messages
for tones.  As others have discussed tones can be also sent via RFC2833, or
as in-band tones. There seems to be a little confusion in some posts over
this so I¹ll touch on it. RFC2833 provides a payload to send tones, other
trunk signals in a RTP media stream where an in-band DTMF signal would be a
digitized tone sent encoded using the codec.

For H.245 when a connection is established the capability to send user input
is confirmed in the terminalCapabilitySet so one end of the connection has
an idea of what the other end can do. I have seen this used for a couple of
reasons, it is predominately used when the media and the signaling follow
different paths, for example signaling through a gatekeeper and media
directly to the endpoints. This allows a client to signal the gatekeeper
without it needing to route and process the media streams. As some have
discussed it also provides a reliable delivery path so that the potential
for loss doesn¹t need to be mitigated by sending multiple tones, etc.

All gateways I have worked with for H.323, SIP, PSTN understand both RFC2833
and H.245(or Q.931) messages and can bounce between them. I haven¹t run into
many that do it in-band within H.323 or SIP because of the issues discussed
earlier in the thread regarding producing reliable tones with lower
bandwidth codecs.

My thought regarding which to support is that it would be good to support a
reliable method as well as RFC2833.

- Brian


On 2/15/06 6:08 PM, "dirk.griffioen at voipster.com" <dgriffioen at voipster.com>
wrote:

> Jean-Louis Seguineau wrote:
>>  
>> No doubt these are definitively serious concerns. But in all fairness, we
>> are here to discuss the merits of two different approaches from a protocol
>> stand point. The people on the Zoep's side have been trying to bring
>> relevant information to the list to allow us to make an informed decision.
>> 
>>   
> Apart from this mailing list, is there a way to sum up pros and cons of both
> approaches?
> 
> Should I start a wiki page and put things in a matrix of some sort?
>>  
>> IMHO, just stating a list of concerns is not fair, if not considered in
>> context, and illustrated with examples. I must be a little stupid, because I
>> cannot figure out how some of the 7 points may apply when taken in the
>> context of Zoep, sorry.
>>   
> I would agree:
>>  
>> Peter, your legitimate objections would gain greater weight and
>> consideration if you were to give examples of where these concerns apply in
>> the context of Zoep. You always advocate examples as very important, andst
>> will be last to deny their usefulness. Without examples, this 7 points list
>> will only resemble common marketing BS you can read here and there. This is
>> not what we want, do we?
>> 
>> And may I also suggest differentiating between the p2p (XMPP only tunnel)
>> and the pc-pstn (XMPP/SIP/POTS gateway) context.
>> 
>> P.S. I also believe this same explanation would have to be done for the
>> Jingle side. But we already have a framework shaping up to compare the two
>> technologies, don't we?
>> 
>> -----Original Message-----
>> Message: 2
>> Date: Mon, 13 Feb 2006 16:35:30 -0700
>> From: Peter Saint-Andre <stpeter at jabber.org> <mailto:stpeter at jabber.org>
>> Subject: Re: [Standards-JIG] Jingle vs. Zoep
>> To: Jabber protocol discussion list <standards-jig at jabber.org>
>> <mailto:standards-jig at jabber.org>
>> Message-ID: <43F117C2.5070906 at jabber.org>
>> <mailto:43F117C2.5070906 at jabber.org>
>> Content-Type: text/plain; charset="iso-8859-1"
>> 
>> -----BEGIN PGP SIGNED MESSAGE-----
>> Hash: SHA1
>> 
>> dirk.griffioen at voipster.com wrote:
>> 
>>   
>>  
>>>  
>>>>  
>>>>>  
>>>>>>  
>>>>>> as
>>>>>> secure as both XMPP and SIP are.
>>>>>>     
>>>>>>           
>>>>>>  
>>>>>  
>>>>  
>>>  
>>> SIP is secure? Authentication is OPTIONAL. From addresses are not
>>> validated and checked. Interdomain communications ("federation") is
>>> still a mess. Sure you can use sips: URIs (forcing TCP and TLS) but most
>>> implementations out there will still use the old sip: URIs (UDP, no
>>> TLS). It's like Jabber in the jabberd 1.0 days (1999-2000) when we
>>> didn't have dialback.
>>> 
>>>   
>>>     
>>>  
>>>>  
>>>> Does jabber then validate 'from' - in a way more than syntactically
>>>> checking if things are ok? Maybe I am missing the point, but why is this
>>>> so important? 
>>>>       
>>>>  
>>>  
>>  
>> 
>> Makes it relatively to do the following:
>> 
>> 1. Send unsolicited communications.
>> 
>> 2. Launch deregistration attacks.
>> 
>> 3. Perform call flooding.
>> 
>> 4. Terminate calls from a third party.
>> 
>> 5. Hijack sessions.
>> 
>> 6. Perform unauthorized call transfers.
>> 
>> 7. Register unauthorized devices.
>> 
>> And yes I consider those fairly serious.
>> 
>> Peter
>> 
>> 
>> 
>>   
> 
> 


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.jabber.org/pipermail/standards/attachments/20060216/f8322720/attachment.html>


More information about the Standards mailing list